In the

    United States Court of Appeals
                 For the Seventh Circuit
                     ____________________


No. 18-2949
KEITH HOGLUND,
                                               Petitioner-Appellant,

                                 v.

RON NEAL, Warden,
                                               Respondent-Appellee.
                     ____________________

         Appeal from the United States District Court for the
         Northern District of Indiana, South Bend Division.
        No. 3:16-CV-313-PPS-MGG — Philip P. Simon, Judge.
                     ____________________

     ARGUED DECEMBER 4, 2019 — DECIDED MAY 14, 2020
                ____________________

   Before MANION, KANNE, and BARRETT, Circuit Judges.
    MANION, Circuit Judge. A jury found Keith Hoglund guilty
of molesting his daughter. The district judge denied his peti-
tion for a writ of habeas corpus. We affirm.
                           I. Overview
    Hoglund married Teresa Malott in 1998. She already had
a 4-year-old son. The marriage produced two children. A.H.
2                                                   No. 18-2949

was born in 1998; her sister in 2001. A.H. testified she twice
tried to tell her mother her father was molesting her. After the
second time, Mallot went to the police. This followed shortly
after Hoglund told her he committed adultery. Detective Hol-
liday interviewed A.H. in February 2006. She said her father
had her perform oral sex on him. So Dr. Butler examined A.H.
in March 2006. Holliday interviewed Hoglund, who denied
the allegations but also made several strange and incriminat-
ing statements. Indiana charged him with child molesting.
A.H. met with Counselor Shestak in 2007 and Dr. Mayle in
2009. At trial in 2010, A.H. testified Hoglund sexually abused
her from the ages of 4 or 5 to about 7. Indiana called Butler,
Shestak, and Mayle to testify. They relayed what A.H. told
them and they essentially said they believed her. Hoglund
also testified. He denied abusing A.H. But the jury found him
guilty. The judge sentenced him to 50 years. After exhausting
state proceedings, he petitioned the district court for a writ of
habeas corpus. He raised two basic issues that survive for us.
    First, he claimed ineffective assistance of counsel because
his trial attorney failed to object properly to hearsay. Defense
counsel made some hearsay objections when the prosecutor
asked the experts to say what A.H. said. But when the prose-
cutor invoked the medical exception under Indiana Rule of
Evidence 803(4), defense counsel failed to assert the lack of a
foundation that A.H. thought she was speaking to the experts
for diagnosis or treatment. The district judge decided defense
counsel was deficient but the state court’s holding that this
did not prejudice Hoglund was not objectively unreasonable.
   Second, he claimed the admission of the experts’ vouching
violated due process. Indiana precedent at the time of trial—
Lawrence v. State, 

(Ind. 1984), overruled by
No. 18-2949                                                     3

Hoglund v. State, 

(Ind. 2012)—allowed lim-
ited, indirect vouching. Some instances of vouching at trial
satisfied this precedent and were admitted. Others did not,
but still came in. On direct appeal, the Indiana Supreme Court
overruled Lawrence and banned indirect vouching. But the
court denied Hoglund relief because the error was harmless.
The district judge questioned the state court’s harmless-error
analysis, but concluded he could not find the determination
that the error did not prejudice Hoglund was unreasonable.
    So the judge denied the petition, but certified the appeala-
bility of these two issues and the issue of whether the due pro-
cess claim was procedurally defaulted. Hoglund appealed.
                            II. Trial
A. A.H.
    A.H. testified her father made her perform oral sex on him
“maybe twice a week, three times a week” from the age of
“[m]aybe 4 or 5” to about her seventh birthday. She gave
graphic, grotesque, extensive, shocking details. She testified
about all five senses, including the taste of her father’s semen:
“Slimy, gooey, disgusting.” She told the jury the acts made
her mouth sore. She testified he showed her pornographic
movies of oral sex. She testified about his manipulation, her
attempts to refuse, and his persistence in making her perform
oral sex. She asked if he was ever going to do this with her
sister because A.H. “didn’t want her to go through it and he
said I don’t know, maybe.” She was very concerned about her
sister. She testified that after the abusive acts her father would
have her eat food to change her breath.
   On cross, defense counsel explored sibling rivalries and
parental favoritism, and attempted to show Hoglund was a
4                                                    No. 18-2949

good, normal family man. A.H. testified she learned from her
mother that her father was cheating and they would divorce.
A.H. was sad and disappointed. Defense counsel subjected
her to extensive, aggressive, probing, even tedious cross, but
her account remained materially consistent and strong.
    On redirect, A.H. testified about the first time she told her
mother about the sexual abuse. She was 5 or 6. Her mother
had her sit in her room until her father came home. When he
did, he talked with A.H. privately. He told her she could not
tell anyone. She also testified about the second time she told
her mother. This time, her mother contacted the police.
B. Dr. Carol J. Butler
    The State called Dr. Butler, a pediatrician. She testified she
saw A.H. in March 2006 at the request of D.C.S. for a sexual
abuse exam, the purpose of which was to interview A.H., do
a physical exam, and provide treatment if needed. The prose-
cutor: “[Y]ou asked her what she was there to see you for,
what did she say?” Defense counsel objected to hearsay. The
prosecutor invoked Indiana Rule of Evidence 803(4)’s excep-
tion for statements made for purpose of medical diagnosis or
treatment. The court overruled the objection. Butler testified
A.H. said she was in for a checkup. Butler continued:
    [S]he told me that her mom [asked] her if her dad was
    hurting her or doing something he shouldn’t be doing
    and she said yes. … She said … her mom asked her …
    because “her dad was cheating on her mom and he was
    tired of her” … .
Butler then relayed A.H.’s account of the abuse. This graphic
hearsay echoed A.H.’s testimony. Butler took a culture for
chlamydia and gonorrhea from A.H.’s throat: negative.
No. 18-2949                                                    5

    The prosecutor asked Butler to indirectly vouch: “[D]o
you believe that she is prone to exaggerate or fabricate sexual
matters?” Defense objected. The prosecutor rephrased the
question: “[D]id you believe that she is prone to exaggerate or
fantasize in sexual matters?” Defense objected again, and
lodged a continuing objection. The court overruled it. Butler
strayed outside Lawrence and directly vouched:
   I don’t believe an eight year old would come into a
   physician’s office to speak about sexual fantasies or
   made up stories. … [F]or an eight year old to come in
   and speak about that in my opinion is not usually a
   fantasy or a story. To be seven or eight and to have this
   knowledge is also not usual. So I believe that what
   [A.H.] told me was the truth because of her age and
   because people don’t—
    Defense objected again. The prosecutor agreed, and tried
to confine the expert to Lawrence. “Do you believe that [A.H.]
was … prone to exaggerate or fantasize?” Butler: “In regards
to what she told me, no.” The court sua sponte struck the com-
ment about whether A.H. was truthful and instructed the jury
to disregard it, but allowed the opinion she was not prone to
exaggerate or fantasize to stand.
C. Teresa Malott
    The State called A.H.’s mother. She testified A.H. and
Hoglund “were extremely close” when A.H. was young. She
testified they were sometimes alone together, sometimes in
the bedroom. “[H]e didn’t like the other two kids would
bother him, he wanted to keep the air conditioning running
in the bedroom, so he would lock the rest of us out.”
6                                                     No. 18-2949

    Malott testified 5-year-old A.H. told her something alarm-
ing: “[S]he came to me about bathing her father in the shower
and other things that I can’t recall at this time or can’t recall at
all which was very alarming.” Malott called Hoglund at work
about what A.H. said. He was upset and came home. He said
it was not true, and would ruin him. Malott believed him, but
stopped leaving A.H. alone with him. He did not bathe her
anymore. But after a while Malott did not notice anything, so
they resumed the routine. Prosecutor: “Did you ever ask
[A.H.] again?” Malott: “Off and on, like twice after that did I
ask her and she would say no.” Defense counsel objected to
hearsay. The court sustained the objection. Malott testified
she did not suspect anything inappropriate was happening.
    Years passed after the first red flag. In November 2005,
Hoglund began working as a truck driver. In January 2006, he
told Malott by phone he had an affair. She was “horrified,”
“[e]xtremely upset and angry.” She decided to divorce. Soon
after this call, Malott asked A.H. if “what she had told me be-
fore was true or not.” A.H. cried. Without saying what A.H.
said, Malott testified her own reaction to what A.H. said was
to load her children and some supplies into a car and leave
because he was returning. She told the police what A.H. said.
D. Christine Ottaviano Shestak
    The State called Shestak, a mental health counselor who
saw A.H. twice in January 2007. Shestak testified A.H. was
referred for anxiety management as trial loomed, but did not
here testify about A.H.’s purposes. The prosecutor asked if
she perceived any indication A.H. “may have fabricated the
story about her abuse out of some need?” Defense counsel ob-
jected but was overruled. Shestak, like Butler, strayed beyond
Lawrence: “Her statements were congruent with her
No. 18-2949                                                    7

experience and I did not see anything that indicated that she
had any need to tell this story.” But the court did not strike
this direct vouching, despite striking Butler’s direct vouching.
    The prosecutor: “[W]hat did [A.H.] tell you during your
interview with her?” Defense objected but said “I understand
the exception.” He did not argue no foundation supported it.
The court overruled the objection. Shestak answered:
   I asked [A.H.] what exactly was the reason that she was
   here. She knew the trial was coming up, I asked her
   what the trial was about and at that point words just
   began to pour out of her really fast describing incidents
   of what she called washing her dad’s penis and as I
   asked further questions I realized she was talking
   about oral sex. She gave me a great deal of detail. …
   Some of the details were that her mouth hurt and she
   would ask her dad to let her stop and he would urge
   her to go on just a little longer.
Shestak related many other graphic, grisly, lurid details A.H.
had told her. These details echoed A.H.’s trial testimony.
    A.H. drew her father’s genital organ for Shestak. The pros-
ecutor offered the drawing into evidence. Defense counsel ob-
jected to hearsay. The prosecutor relied on the medical excep-
tion, defense counsel failed to challenge foundation, the court
overruled the objection, and the jury saw the drawing.
   Shestak shed some light on A.H.’s early attempt around
age 5 to tell her mother about the abuse:
   Her worst event was being called a liar and … she told
   me that that was when she had told her mother a cou-
   ple years earlier before the final disclosure that, about
   the incident of oral sex with dad that mom had gone to
8                                                   No. 18-2949

    dad, dad had gotten very angry and then she was
    afraid of dad’s anger so she then told her mother that
    it hadn’t happened. She was then grounded for lying
    and that was her worst memory.
    After yet more lurid, revolting hearsay, echoing A.H.’s tes-
timony, the prosecutor asked: “[D]o you believe that she is
prone to exaggerate or fantasize in sexual matters?” Shestak:
“I did not feel there was any great exaggeration. I felt like she
really didn’t want to talk, but she felt like she had to.” The
prosecutor pressed on: “Did you learn anything from your in-
terviews with the child … which would be inconsistent with
[A.H.] being the victim of a sexual abuse?” No objection.
Shestak: “No, I did not.”
E. Detective Scott Holliday
   The State called Holliday, who investigated the allegations
and interviewed A.H. and Hoglund. He testified about in-
criminating statements Hoglund made:
    I told him that his daughter … made allegations that
    he put stuff on his penis and have her lick it off. … I
    believe his response was “No way”. … I did ask him
    why he thought [she] would make such a statement. …
    [H]e [referred] to an incident … a few years back …
    where he had left a pornographic movie in the VCR …
    the next day … he came in from the garage and I be-
    lieve what he told me he was horrified to see [she] was
    watching, his words “an oral sex movie” … .
    Holliday also testified Hoglund said A.H. walked in on
him and his wife having oral sex several times, but Hoglund
could not give details. Holliday also testified Hoglund
claimed A.H. walked in on him masturbating and ejaculating.
No. 18-2949                                                       9

Holliday testified he asked if there was ever a time his daugh-
ter might have put her mouth on his genital organ:
   [W]hen I asked … if there was ever a time when she
   put her mouth on his penis, he had made comment that
   not unless [he] was passed out sleeping and [she] took
   it upon herself. … He had also told me that there had
   never been a time … where he was coherent that it hap-
   pened. … [H]e had also made statement to that same
   question that he had been trying to think back over
   time, but couldn’t recall this happening, he had also
   made comment that he was trying to think if it could
   have happened when he was f*cked1 up. … His words.
    Holliday was present when A.H. testified. The prosecutor
asked him if her testimony was consistent with what she said
at the interview in February 2006. He said it was. 2
F. Dr. Amanda Mayle
     The State called Dr. Mayle, a clinical psychologist. She tes-
tified she saw A.H. in July and August 2009 because the pros-
ecutor asked for an assessment of emotional stability and abil-
ity to testify. Mayle found her able to testify. The prosecutor
asked Mayle to say what A.H. said. This time, defense counsel
did not object to hearsay, though no foundation supported
the exception. Mayle testified:
   [A.H.] stated … the abuse start[ed] when she was
   around 4 or 5 [and] ended when she was close to 8. She

   1   The trial transcript, of course, spells out the word.
   2 Defense counsel made no objection here. Indeed, defense counsel
had the witness reiterate the comparison on cross examination. And
Hoglund raises no quarrel on this point to us.
10                                                           No. 18-2949

     said “He made me suck his privates.” She also said …
     her dad had a red box with little bottles that were dif-
     ferent flavors [and] he would wipe this on his privates
     before he would make her suck his privates. … She
     said afterward he would give her money and some-
     thing to eat like a brownie to change the smell of her
     mouth. … She also stated that he would tell her not to
     tell because “Daddy would go to jail and you wouldn’t
     want that.” She also stated that he would often grab
     her neck while she was sucking his privates and that
     she to this day hates when people touch her neck and
     that it kind of makes her have triggers of past events.
     She also stated that she asked him if he would ever do
     this to her little sister and he stated maybe. She said
     that made her very angry, she didn’t like that. Stated
     that she … told her sister during a secrets game and
     that she had told her mom too and that … her mom
     had talked to her dad about it and that he had pulled
     her in her room later and told her not to tell and threat-
     ened her and told her … they were still going to do it
     and she told him that she didn’t want to because she
     didn’t like to do it anymore and … he had smacked her
     in the face and left a red mark. She also stated that she
     was afraid because she thought … one of dad’s friends
     tried to run her over with a red van … .
  Mayle relayed what she says A.H. said, echoing her testi-
mony and adding new details.3 The prosecutor asked: “[D]o

     3A.H. did not testify about her father grabbing her neck or smacking
her face, telling her sister about the abuse, or a red van; no one asked her
about these details at trial. On appeal to us, Hoglund does not raise a par-
ticular challenge to these details.
No. 18-2949                                                   11

you perceive any indication that [A.H.] may have fabricated
this story of her abuse out of some need?” Over an overruled
objection, Mayle said no. The prosecutor asked if she believed
A.H. was prone to exaggerate or fantasize in sexual matters.
Mayle said no. The prosecutor nudged Mayle further: “Did
you learn anything about [A.H.] which you believe would be
inconsistent with the victim being a victim of sexual abuse?”
No objection. Mayle said no. The State rested.
G. Hoglund
    Hoglund took the stand. He testified about activities with
his children, his work, and his decision to be a truck driver.
He left for the road in November 2005. He returned Christmas
Day with gifts. He left the next day. He testified he cheated.
He told his wife on the phone in January 2006. “[S]he was ex-
tremely pissed off, screaming at me on the phone, called me
every name in the book … .” He testified about an incident
when he claims he accidentally left a pornographic movie fea-
turing oral sex in the VCR and A.H. watched it:
   [M]y Friday nights I usually went out blowing off
   steam from work and I came home, everybody was in
   bed and we had these x-rated movies and I just popped
   one in and I started watching it and you know I took
   care of what I needed to take care of and then I went
   off to bed. The next day I’m out in my garage … and
   my wife … came out and said she was going to the
   store and I was like yeah, okay, whatever, where are
   the kids and she was like oh they are in the yard … . So
   I went about my business … . [T]hen I thought well I
   better go make sure everything is alright and I needed
   to get a drink anyway, so I walked into the house to
   my horror there’s my little girl watching this movie
12                                                        No. 18-2949

     that I had left in the VCR. … I basically wanted to kick
     my ass for leaving it in there, pardon my language. I
     was horrified, I was lowered … I would never, never
     bring my daughter in to watch a movie like that.
   Hoglund also testified about an incident when he claims
A.H. saw him ejaculate:
     The wife and kids were at the store, nobody was home
     and I thought it would be a perfect opportunity to re-
     lieve myself and low and behold I didn’t hear them
     coming in, in walks in [A.H.] and you know I turned
     away, I told her you need to get out of here, basically I
     got caught.
He also testified about an incident when he claims A.H. saw
him and his wife engaging in oral sex.
   He testified that when Holliday confronted him with
A.H.’s accusations, he “was devastated … instantly angry,
shocked, couldn’t understand where these allegations would
come from.” He gave explanations for his strange state-
ments. 4


4 Hoglund: “Holliday … asked me if I thought at one point if I might have
been messed up enough to do something like this with my kids.”
Defense counsel: “How did you respond to that?”
Hoglund: “Never. … I couldn’t do that to a kid.”
Counsel: “Also there were two statements … made by you and one of
them … was that my daughter knows more about sex than we do, how
was that statement made by you at the time?”
Hoglund: “I was mad, I was frustrated, I had just been delivered a bomb
so to speak. It was a bit of an exaggeration.”
Counsel: “There was another question … which your response … was
‘Unless I was passed out and she on her own would have done something
like that’ … . Would you please explain that response?”
No. 18-2949                                                              13

   Hoglund unequivocally denied the allegations. Defense
counsel asked how he felt hearing A.H. say she put her mouth
on his genital organ. “I was hurt, I was stunned, I wondered
how my baby girl could say this about me, what did I ever
do.” Defense counsel asked if he ever had A.H. put her mouth
on his genital organ and put flavoring on it. “I never did. I
never would, I find that disgusting, appalling and that’s my
baby girl. I mean I use to feed her, change her diapers, she was
my first baby. I would never hurt my kids. I swear to God I
wouldn’t.” Defense rested. The jury convicted.
                      III. Further Proceedings
A. Direct Appeal to Indiana Court of Appeals
    Hoglund appealed, challenging the vouching by the ex-
perts. In its decision, the Indiana Court of Appeals quoted the
experts’ testimony at length, including several passages of di-
rect or arguably direct vouching. Here are highlights:
Butler: “So I believe that what [A.H.] told me was the truth
because of her age and because people don’t—”
Shestak: “Her statements were congruent with her experi-
ence … .”


Hoglund: “[M]y mind was spinning in so many different directions from
absorbing the information [he] was giving to me and he was explaining …
the things that he had seen in his career … and … that was basically an
exaggeration … like my daughter knowing about as much sex as … an
adult … . I felt like he was trying to say well what about on a subconscious
level, you know no, that would never happen consciously, unconsciously,
drunk, stoned, whatever I would never, ever do that to my daughter or
any child for that matter. … I was stunned basically the whole time I talked
to him. I don’t know how to explain it, but I have never been faced with
anything in of that nature and to hear something like that … my mind just
went 100 miles a minute, numb I guess.”
14                                                   No. 18-2949

Mayle: “No [I did not perceive any indication A.H. may have
fabricated this story of her abuse out of some need.] … No [I
did not learn anything about A.H. which I believe would be
inconsistent with a victim being a victim of child abuse].”
    Yet the appellate decision suggested Hoglund was not
challenging any direct vouching: “In any event, Hoglund
does not dispute that the evidence at issue here is indirect
vouching by an expert under Lawrence. He argues, however,
that Lawrence is no longer good law.” Hoglund v. State, 

, 171 (Ind. Ct. App. 2011). But the Indiana Court of
Appeals found itself bound by Lawrence and concluded the
trial court did not abuse its discretion in admitting indirect
vouching for A.H. The concurrence expressed concern about
the direct vouching and the avalanche of accumulating
vouching. But the concurrence recognized “there is no entitle-
ment to a perfect trial.”

(Darden, J., concurring) (in-
ternal quotation marks omitted). The concurrence concluded
the vouching was not so prejudicial as to require reversal,
given the “articulate and detailed testimony of A.H.” and
given “Hoglund’s own testimony as to how A.H. would not
have been aware of such details without having personally
experienced them with him … .”

Appeal to Indiana Supreme Court
    At the next step, the Indiana Supreme Court granted
Hoglund a hollow victory by overruling Lawrence but still af-
firming. The court ruled testimony that an alleged child vic-
tim is not prone to exaggerate or fantasize in sexual matters is
an indirect but functional equivalent of saying the child is tell-
ing the truth. Thus the court overruled Lawrence and joined
the majority of States. The court held the trial court erred in
allowing vouching into evidence over Hoglund’s objection.
No. 18-2949                                                    15

But the court concluded the improper admissions were harm-
less errors because there was substantial evidence of guilt
apart from the erroneously admitted vouching: “A.H. testi-
fied at length concerning what happened to her at the hands
of her father. And her testimony remained consistent and un-
shaken under aggressive cross examination.” Hoglund v. State,

, 1238 (Ind. 2012). Also, the court found the
erroneously admitted-over-objection vouching to be cumula-
tive of other vouching admitted without objection.
C. State Petition for Post-Conviction Relief
    Hoglund petitioned for post-conviction relief and lost. On
appeal, he argued his trial counsel rendered ineffective assis-
tance by failing to object to all expert vouching testimony, fail-
ing to seek a mistrial because of direct and overt vouching by
Butler, and failing to object properly to hearsay. The Indiana
Court of Appeals correctly recited the Strickland test, as re-
stated by the Indiana Supreme Court: “To prevail on a claim
of ineffective assistance of counsel, a petitioner must demon-
strate both that his counsel’s performance was deficient and
that the petitioner was prejudiced by the deficient perfor-
mance. A counsel’s performance is deficient if it falls below
an objective standard of reasonableness based on prevailing
professional norms. To meet the appropriate test for preju-
dice, the petitioner must show that there is a reasonable prob-
ability that, but for counsel’s unprofessional errors, the result
of the proceeding would have been different.” Hoglund v.
State, 

, No. 90A02-1503-PC-182, 

, at *2 (Ind. Ct. App. Feb. 5, 2016) (internal citations
omitted), transfer denied, 

(Ind. 2016).
   The court first turned to Hoglund’s claim that trial counsel
performed deficiently by failing to object consistently to
16                                                   No. 18-2949

vouching. Trial counsel objected to some vouching, but not
all. But the court observed that at the time of trial, Lawrence
permitted indirect vouching, so the trial court would have
overruled even consistent and thorough objections to indirect
vouching, so the failure to make consistent and thorough ob-
jections to indirect vouching was not ineffective assistance.
    The Indiana Court of Appeals then turned to Hoglund’s
argument that trial counsel was deficient by failing to seek a
mistrial after Butler directly vouched for A.H. by testifying, “I
believe that what [A.H.] told me was the truth because of her
age … .” Trial counsel lodged an objection here, and the pros-
ecutor even agreed on record. The trial court struck the testi-
mony and admonished the jury to disregard it. The Court of
Appeals concluded the trial court timely and properly ad-
monished the jury, which presumably cured any error regard-
ing the testimony. The Court of Appeals also concluded the
trial court would have denied any request for a mistrial at this
point, as mistrial is an extreme remedy granted only when
nothing else can rectify the situation. Thus, trial counsel’s per-
formance on this point was deemed effective.
    Finally, the court turned to Hoglund’s claim that trial
counsel failed to object properly to the hearsay testimony of
Mayle and Shestak. Trial counsel objected several times to the
hearsay, the prosecutor claimed admissibility under 803(4)’s
exception for statements made for the purpose of medical di-
agnosis or treatment, but trial counsel failed to assert the lack
of a foundation for that exception, so the judge admitted the
testimony. The appellate court agreed trial counsel’s perfor-
mance was deficient in failing to object to the lack of a foun-
dation for Mayle’s and Shestak’s hearsay testimony, but con-
cluded any error in the admission was harmless given A.H.’s
No. 18-2949                                                    17

testimony, the aggressive cross of A.H., and the fact that the
hearsay testimony was “merely cumulative” of A.H.’s testi-
mony. Thus, the court deemed trial counsel not ineffective.
   The Indiana Supreme Court denied transfer.
D. Federal Petition for Writ of Habeas Corpus
    Hoglund petitioned under 28 U.S.C. § 2254 for habeas cor-
pus relief. The only claims remaining before us are the claim
that the trial court violated due process by admitting expert
testimony vouching for A.H. and the claim that trial counsel
rendered ineffective assistance by failing to object to expert
testimony on the basis of hearsay and lack of foundation for
the medical exception. In a thoughtful and thorough opinion,
the district judge determined Hoglund procedurally de-
faulted the due process / vouching claim because he relied
only on state cases and evidentiary rules when presenting this
to the state courts, so he failed to alert them to the federal na-
ture of his claim. But the judge decided to consider the merits
of whether the admission of vouching violated due process
because this claim is closely related to one properly ex-
hausted. The judge determined Hoglund properly exhausted
his state remedies for his claims that trial counsel failed to ob-
ject to expert testimony on the basis of hearsay and lack of
foundation, that trial counsel failed to object to expert testi-
mony that vouched for A.H., and that trial counsel failed to
request a mistrial after overt and direct vouching.
    The judge turned to the merits of Hoglund’s claim that
trial counsel was ineffective by failing to properly object to
hearsay admitted under the medical exception without the
proper foundation. The judge correctly recited the Strickland
standard: prejudice caused by deficient performance.
18                                                 No. 18-2949

Prejudice is a reasonable probability that but for the deficien-
cies the result would have differed. A reasonable probability
is a probability sufficient to undermine confidence in the re-
sult. On habeas, there is double deference because the inquiry
is whether the state court unreasonably applied Strickland.
    Trial counsel objected to Butler and Shestak relaying what
A.H. told them, but failed to argue the exception did not ap-
ply. He did not object at all to Mayle’s hearsay testimony, per-
haps because he already lost on this issue. At the post-convic-
tion relief stage, the Indiana Court of Appeals held trial coun-
sel performed deficiently by failing to object to hearsay and
lack of foundation, but the objectionable testimony did not
prejudice Hoglund. The district judge agreed counsel per-
formed deficiently. But the judge found the issue of whether
this deficiency prejudiced Hoglund to be a close question.
    The judge granted that the improperly challenged hearsay
testimony “unquestionably bolstered the credibility of the
victim making her testimony much more believable” and “a
strong argument could be made that the outcome of the trial
may very well have been different without this added boost
to the victim’s credibility.” But the judge recognized that
A.H.’s testimony was compelling, graphic, consistent, and de-
tailed, and that a jury could have convicted based on this tes-
timony alone. Also, Hoglund’s own statements to the police,
although not a confession, were highly incriminating. Finally,
Mallot testified about a bathing episode involving Hoglund
and then 5-year-old A.H. So the judge concluded the case
against Hoglund “was strong—not overwhelming, but
strong—even without the testimony of the three experts.” The
judge could not find that the state court’s determination that
the error did not prejudice Hoglund was unreasonable.
No. 18-2949                                                    19

    The judge went on to say that even if trial counsel had
properly objected for lack of foundation for the medical ex-
ception, the prosecutor might have been able to lay a founda-
tion for at least Butler’s hearsay testimony because it appears
Butler was actually treating A.H. (Of course, the prosecution
would have had to establish not merely that Butler was actu-
ally treating A.H., but that A.H. understood the purpose of
her visit to Butler at the time.)
    The judge turned to the issue of vouching. Hoglund
claimed it was objectively unreasonable for the state court to
determine that trial counsel’s failure to continuously object to
vouching and failure to request a mistrial after direct and
overt vouching did not constitute deficient performance.
Hoglund also claimed the admission of vouching violated his
due process right to a fundamentally fair trial. The judge
noted the trial was “replete with vouching” and quoted many
passages at length from the testimony of all three experts. The
judge then observed a dichotomy in the state proceedings. On
direct appeal, the Supreme Court overruled Lawrence and
held that vouching testimony in sexual abuse cases was no
longer admissible, but affirmed the trial court’s judgment be-
cause the admission of the vouching was harmless error given
A.H.’s compelling testimony and given that the erroneously
admitted vouching was cumulative of other vouching admit-
ted without objection. So at the post-conviction relief stage,
Hoglund argued trial counsel was ineffective for failing to ob-
ject continuously to vouching by Mayle and Shestak and for
failing to seek a mistrial after Butler testified A.H. was telling
the truth. But the Court of Appeals reasoned trial counsel’s
performance was not deficient because the Supreme Court
had not yet changed the evidentiary rule by the time of trial
and because the striking and admonition solved the problem
20                                                   No. 18-2949

short of a mistrial. So, “on direct appeal, Hoglund was told
that the trial court error was harmless due to trial counsel’s
mistakes; but, on post-conviction appeal, Hoglund was told
that trial counsel’s performance was not deficient; and neither
decision discussed whether the vouching testimony in its en-
tirety was prejudicial.”
   Though the district judge questioned the state court’s
harmless-error analysis, he recognized that the issue before
him was whether the vouching testimony produced a signifi-
cant likelihood an innocent person has been convicted. He
could not reach that conclusion, given A.H.’s “extremely com-
pelling” testimony and given Hoglund’s strange and incrim-
inating statements to the police.
    The judge granted a certificate of appealability on three is-
sues: 1) whether trial counsel’s failure to object properly to the
hearsay from the three experts relaying what A.H. had told
them constitutes ineffective assistance of counsel; 2) whether
the admission of vouching testimony was a due process vio-
lation; and 3) whether Hoglund procedurally defaulted the
due process claim on the vouching testimony.
                         IV. Standards
    We review the district judge’s denial of habeas corpus re-
lief de novo. The Antiterrorism and Effective Death Penalty
Act limits a federal court’s ability to grant a state prisoner’s
petition for a writ of habeas corpus:
     (d) An application for a writ of habeas corpus on behalf
     of a person in custody pursuant to the judgment of a
     State court shall not be granted with respect to any
     claim that was adjudicated on the merits in State court
     proceedings unless the adjudication of the claim—
No. 18-2949                                                   21

   (1) resulted in a decision that was contrary to, or in-
   volved an unreasonable application of, clearly estab-
   lished Federal law, as determined by the Supreme
   Court of the United States; or
   (2) resulted in a decision that was based on an unrea-
   sonable determination of the facts in light of the evi-
   dence presented in the State court proceeding.
28 U.S.C. § 2254(d). This standard is “difficult to meet” and
“highly deferential.” Cullen v. Pinholster, 

, 181
(2011). It is not enough for a petitioner to show the state
court’s application of federal law was incorrect; rather, he
must show the application was unreasonable, which is a “sub-
stantially higher threshold.” Schriro v. Landrigan, 

,
473 (2007). A petitioner “must show that the state court’s rul-
ing on the claim being presented in federal court was so lack-
ing in justification that there was an error well understood
and comprehended in existing law beyond any possibility for
fair-minded disagreement.” Harrington v. Richter, 

,
103 (2011). A state court’s decision can be a reasonable appli-
cation of Supreme Court precedent even if we think it is an
incorrect application, and even if petitioner presents a strong
case for relief. Subdiaz-Osorio v. Humphreys, 

, 443
(7th Cir. 2020). This “deferential standard of review applies
only to claims that were actually ‘adjudicated on the merits in
State court proceedings.’” Campbell v. Reardon, 

,
762 (7th Cir. 2015) (quoting § 2254(d)). For each claim, “we
review the decision of the last state court to address its mer-
its.” Snow v. Pfister, 

, 864 (7th Cir. 2018).
   AEDPA generally requires a state prisoner to exhaust the
remedies available in state court. 28 U.S.C. § 2254(b)(1)(A). Be-
fore bringing a claim on federal habeas, a state prisoner must
22                                                            No. 18-2949

present the claim to the state courts so they have a “fair op-
portunity” to consider and correct the alleged problem. Picard
v. Connor, 

, 275–76 (1971). Fair presentment “does
not require a hypertechnical congruence between the claims
made in the federal and state courts; it merely requires that
the factual and legal substance remain the same.” Anderson v.
Benik, 

, 814–15 (7th Cir. 2006). The basic question
is whether “in concrete, practical terms … the state court was
sufficiently alerted to the federal constitutional nature of the
issue to permit it to resolve that issue on a federal basis.” Kur-
zawa v. Jordan, 

, 442 (7th Cir 1998). We give a “gen-
erous interpretation” to pro se filings in this context. Lewis v.
Sternes, 

, 1027 (7th Cir. 2004).
   Hoglund brings two claims to us: (1) his trial counsel ren-
dered ineffective assistance by failing to object properly to
hearsay testimony from the experts relaying what A.H. told
them before trial; and (2) the admission of the experts’ vouch-
ing testimony violated due process.
       V. Ineffective Assistance of Counsel on Hearsay
   Hoglund claims trial counsel rendered ineffective assis-
tance by failing to argue no foundation supported the medical
exception for the experts to relay what A.H. told them. 5


5 A bit of housekeeping about Butler. In his brief before us, Hoglund
lumps all three experts—Butler, Shestak, and Mayle—into this challenge.
And the district judge seemed to consider all three regarding this chal-
lenge. He determined the prosecution might have been able to establish a
foundation for at least one of them: Butler. The district judge certified the
appealability of this issue regarding all three. But the state post-conviction
Court of Appeals noted Hoglund only claimed his trial counsel failed to
object properly to the hearsay testimony of two experts: Shestak and
Mayle. For the proposition in his opening appellate brief that on state post-
No. 18-2949                                                               23

A. Standard
     To succeed on a claim of ineffective assistance of counsel,
petitioner must show (1) “counsel made errors so serious that
counsel was not functioning as the ‘counsel’ guaranteed the
defendant by the Sixth Amendment” and (2) the errors de-
prived the defendant of a fair trial. Strickland v. Washington,

, 687 (1984). To satisfy the second prong, peti-
tioner must show there is a “reasonable probability that, but
for counsel’s unprofessional errors, the result of the proceed-
ing would have been different.”

This “reasonable
probability” is “a probability sufficient to undermine confi-
dence in the outcome.”

whether there is a rea-
sonable probability that the errors changed the outcome of the
trial, the court must consider all of the evidence.” Cook v. Fos-
ter, 

, 909 (7th Cir. 2020).
    When asserting an ineffective assistance of counsel claim
in an AEDPA petition for a writ of habeas corpus, petitioner
faces the obstacle of double deference. He must show the state
court’s decision is contrary to or involved an unreasonable

conviction review he argued trial counsel failed to appropriately object to
hearsay testimony by the experts, Hoglund only cites the state post-con-
viction Court of Appeals’ decision. In its federal appellate response, the
government notes that in the state courts, Hoglund only claimed trial
counsel was deficient for not making better objections to Mayle and
Shestak, and Hoglund does not contest this in his reply. So there might be
an exhaustion problem regarding Butler. But resolution is unnecessary.
Whether or not the challenge to Butler’s hearsay testimony is properly be-
fore us, it is likely the prosecution established (or could have established)
a foundation for the medical exception for this testimony. And more to the
point, either way, the strong, graphic, consistent testimony of A.H. (even
under aggressive and at times tedious cross) coupled with Hoglund’s
strange and incriminating statements assure us trial counsel’s deficient
803(4) performance did not prejudice Hoglund.
24                                                 No. 18-2949

application of the Strickland standard. The last Indiana court
to address the merits of Hoglund’s ineffective assistance of
counsel / hearsay claim was the post-conviction Court of Ap-
peals in 2016. So we review that opinion. The Court of Ap-
peals concluded trial counsel performed deficiently by failing
to object to the lack of a foundation for the admission of hear-
say by Mayle and Shestak under the medical exception. But
the court determined any error here was harmless given
A.H.’s testimony, the aggressive cross of A.H., and the fact
that the hearsay testimony was “merely cumulative” of A.H.’s
testimony. So there was no prejudice. Hoglund, 

, at *6.
B. Deficient Performance?
    The state post-conviction Court of Appeals, the district
judge, and both parties on appeal to us seem to agree trial
counsel was deficient by failing to object to the lack of foun-
dation for the medical hearsay exception. At trial, Rule 803(4)
provided a hearsay exception for statements made for medi-
cal diagnosis or treatment, provided a foundation was laid:
“Statements made for purposes of medical diagnosis or treat-
ment and describing medical history, or past or present symp-
toms, pain, or sensations, or the inception or general character
of the cause or external source thereof insofar as reasonably
pertinent to diagnosis or treatment.” Rule 803(4) (effective
1994).
   The point of this exception is that a court can derive some
assurance of the hearsay statement’s credibility from the fact
that it was made for medical diagnosis or treatment because
an out-of-court declarant generally has a strong self-interest
in making truthful statements to a medical provider when
seeking diagnosis or treatment lest the medical provider
No. 18-2949                                                   25

misdiagnose and treat for allergies instead of the plague. But
for this exception to serve its purpose, there must be some ba-
sis to think the speaker understood the medical purpose of the
statement. See McClain v. State, 

, 331 (Ind. 1996).
Thus, Indiana courts have recognized that alleged child vic-
tims might be too young for a fair presumption they under-
stood the medical purpose, and have required a foundation
that they had this understanding. See

infer-
ence is not obvious, as in this case involving a young child
brought to treatment by someone else, there must be evidence
that the declarant understood the professional’s role in order
to trigger the motivation to provide truthful information.”).
    The testimony of Mayle (the third expert to testify) is the
clearest example of the problem. A.H. gave no testimony
about why she thought she was visiting Mayle. Mayle testi-
fied she saw A.H. because the prosecutor’s office asked for an
assessment of her emotional stability and ability to testify.
Mayle offered no testimony about A.H.’s understanding of
the reason for the visits. No evidence showed A.H. thought
she sought medical diagnosis or treatment from Mayle. But
when the prosecutor asked Mayle to testify about what A.H.
said, defense counsel did not object to hearsay at all, perhaps
because he already lost his hearsay objections to the prior ex-
perts. Failing to object properly to hearsay was deficient. It
allowed Mayle to testify at length about what A.H. said, ech-
oing A.H.’s testimony, thereby lending credibility to A.H.
(And it went beyond A.H.’s testimony.)
    The testimony of Shestak (the second expert to testify) was
also problematic. A.H. gave no testimony about why she
thought she was visiting Shestak. Shestak testified A.H. was
referred to her for help managing anxiety she experienced as
26                                                 No. 18-2949

trial approached, which might have hinted at the possibility
for the medical exception to hearsay. But Shestak’s testimony
about A.H.’s understanding of the purpose of the visits did
not establish the requisite foundation: “I asked [A.H.] what
exactly was the reason that she was here. She knew the trial
was coming up, I asked her what the trial was about and at
that point words just began to pour out of her really fast de-
scribing incidents … .” The prosecutor asked Shestak to relay
what A.H. told her. Defense counsel objected but folded,
failed to challenge foundation, and simply said “I understand
the exception.” The court overruled the objection. Shestak re-
layed the graphic, lurid details A.H. gave. Defense counsel
performed deficiently by failing to challenge the foundation
for the hearsay exception during Shestak’s testimony.
    Butler was the first expert to testify. A.H. did not testify
about why she thought she was visiting Butler. Butler testified
she saw A.H. at the request of D.C.S. for a sexual abuse exam,
and the purpose of this exam included medical diagnosis and
treatment. The prosecutor asked Butler to relay what A.H.
said about the purpose. Defense counsel objected to hearsay.
The prosecutor invoked the 803(4) exception. Defense counsel
failed to argue foundation. (The question, after all, went to
foundation.) The court overruled the objection. Butler testi-
fied A.H. said she was in for a checkup. This might be, or at
least approaches, a satisfactory foundation for the medical ex-
ception. “Checkup” is commonly used by children and adults
to mean a medical visit for diagnosis or treatment. It is possi-
ble, perhaps even likely, arguments attacking foundation here
would have failed. We do not know with certainty because
the issue was not developed. Defense counsel did not chal-
lenge foundation, so the prosecution had no occasion to elab-
orate. There was no voir dire or other further inquiry on this
No. 18-2949                                                    27

point. Butler proceeded to relay A.H.’s graphic account of the
sexual abuse, echoing the testimony A.H. just gave the jury.
   Defense counsel’s performance regarding the hearsay tes-
timony of at least Mayle and Shestak was deficient.
C. Prejudice?
    The next question for the post-conviction Court of Ap-
peals was whether this deficient performance prejudiced
Hoglund. The court held it did not because A.H. testified and
“was aggressively cross-examined.” Her testimony mirrored
the hearsay given by Mayle and Shestak, “making the expert
witnesses’ testimony merely cumulative and, at most, harm-
less error.” Hoglund argues the post-conviction Court of Ap-
peals’ decision on this point was contrary to or involved an
unreasonable application of clearly established federal law, as
determined by the Supreme Court of the United States.
1. contrary to clearly established federal law?
   Hoglund argues the state court’s decision was contrary to
federal law because the court applied the wrong standard.
Shortly before presenting its conclusion, the court quoted
VanPatten v. State, 

, 267 (Ind. 2013), for the
proposition that “[a]dmission of hearsay evidence is not
grounds for reversal where it is merely cumulative of other
evidence admitted.” Hoglund argues this is the wrong stand-
ard. He argues the appropriate question was not whether ad-
mission of hearsay was sufficiently prejudicial to warrant re-
versal of the conviction but merely whether there was a rea-
sonable probability that, absent the errors, the jury would
have had a reasonable doubt about guilt. He argues Indiana
courts use varying terminology to define reversible error. He
points to Jaske v. State, 

, 22 (Ind. 1989), where the
28                                                 No. 18-2949

reversible-error inquiry was whether there is a “substantial
likelihood” that the challenged evidence contributed to the
conviction. He points to Miller v. State, 

, 1114
(Ind. 1982), for the proposition that the reversible-error in-
quiry was whether the error had “substantial influence.”
Hoglund argues the correct standard here requires only a rea-
sonable probability of a different outcome, not the substantial
likelihood of a different outcome required by the state court,
so the state court’s analysis was contrary to Strickland.
   A state court’s decision is contrary to clearly established
federal law for these purposes if the state court’s decision is
“diametrically different, opposite in character or nature, or
mutually opposed” to Supreme Court precedent. Williams v.
Taylor, 

, 405 (2000) (internal quotation marks
omitted). There are several potential reasons to think the state
court here did not apply a standard contrary to Strickland.
    First, the state court correctly discussed the Strickland
standard earlier in its opinion. The court observed Strickland
requires prejudice caused by deficient performance. The court
wrote: “To meet the appropriate test for prejudice, the peti-
tioner must show that there is a reasonable probability that,
but for counsel’s unprofessional errors, the result of the pro-
ceeding would have been different.” Hoglund complains this
was eight pages before the conclusion, but that is only because
the court discussed three other potential areas of ineffective
assistance of counsel before reaching hearsay.
    Second, the actual analysis the state court gave—that A.H.
testified and was aggressively cross examined, and the hear-
say testimony was merely cumulative of A.H.’s testimony—
is not contrary to (or an unreasonable application of) Strick-
land. The court said the problematic hearsay testimony was
No. 18-2949                                                   29

merely cumulative of A.H.’s aggressively cross-examined tes-
timony, and was at worst harmless. That is, trial counsel’s de-
ficient performance did not prejudice Hoglund because there
is no reasonable probability of a different outcome without
the deficiency.
    Third, we are not convinced the state court’s quotation of
VanPatten’s language about “reversal” suggests the gap
Hoglund claims. After all, the state court also quoted lan-
guage from VanPatten that “errors in the admission of evi-
dence are to be disregarded as harmless error unless they af-
fect the substantial rights of a party.” If an error is harmless
and does not affect the substantial rights of a party, then it
does not cause the party Strickland prejudice. If an error truly
results in merely cumulative, harmless repetition, then it does
not prejudice the party. And the court did not cite or use “sub-
stantial likelihood” language. Hoglund imported that.
    Ultimately, it is unclear whether the state court applied the
wrong standard or whether “it is more likely that the court
stated its conclusion imprecisely than that it applied a differ-
ent standard,” as we held in Malone v. Wells, 

, 758
(7th Cir. 2008). But we need not decide that question because
even under a de novo application of Strickland, without
AEDPA deference, we conclude trial counsel’s deficient per-
formance did not prejudice Hoglund.
2. unreasonable application of clearly established federal law?
    Because he claims the state court applied the wrong stand-
ard, Hoglund asks us to independently determine whether
there was a reasonable probability the outcome would have
differed. There was not. We independently conclude there
was no reasonable probability that, but for counsel’s deficient
30                                                 No. 18-2949

performance on hearsay, the result of the trial would have dif-
fered.
    A.H.’s testimony was strong, graphic, detailed, and inter-
nally consistent. When A.H. did not remember something,
she said so. A.H. testified about a course of sexual abuse over
an extended period of time. As noted, her testimony included
all five senses. One might say her testimony even included a
sixth-sense premonition her father might consider abusing
her younger sister. Hoglund’s trial counsel subjected her to
extensive, probing, aggressive, detailed, even at times tedious
cross examination, but she held up. Her account remained
consistent. A jury may convict on the basis of the alleged child
victim’s testimony alone. Deaton v. State, 

, 456
(Ind. Ct. App. 2013) (collecting cases). We agree with the dis-
trict judge’s assessment that “the victim’s testimony was com-
pelling with respect to the graphic nature of the conduct de-
scribed and its consistency and detail. A reasonable jury could
have convicted on the basis of this testimony alone.”
    Hoglund flatly denied abusing his daughter, but he also
made multiple strange and incriminating statements. Malott
testified about opportunity. A.H. and Hoglund were ex-
tremely close and spent time alone together, including in a
locked bedroom. Malott testified about A.H. reporting abuse
twice, and about fleeing the house with her children. Holliday
testified Hoglund denied the accusations: “No way.” But
Hoglund also made strange statements. He talked about A.H.
watching an oral sex movie he left in the VCR. He talked
about A.H. seeing him and his wife having oral sex. He talked
about A.H. seeing him masturbating and ejaculating, and
Holliday quoted Hoglund as saying during the interrogation:
“‘The f*cking sh*t went everywhere.’” After prodding by
No. 18-2949                                                            31

Holliday, Hoglund “made comment that not unless [he] was
passed out sleeping and [she] took it upon herself.” Hoglund
claimed there was never a time when he was coherent that it
happened. He said he was trying to think if it could have hap-
pened when he was “f*cked up.” Holliday testified A.H.’s tes-
timony was consistent with what she told him years earlier.
   Even without the objectionable hearsay, the case against
Hoglund was strong. The evidence was not in equipoise, as
Hoglund claims. Trial counsel’s deficient performance did
not prejudice Hoglund. There is no reasonable probability
that, but for counsel’s deficient performance on hearsay, the
result of the trial would have been different. 6
                 VI. Due Process and Vouching
   Hoglund also claims entitlement to habeas relief because
the admission of improper vouching testimony violated his
right to due process.
A. Procedural Default?
    First, we must address whether Hoglund procedurally de-
faulted this claim. Appellee argues Hoglund failed to raise a
federal due process claim to the Indiana courts, so this claim
is barred. According to Appellee, Hoglund based his direct-
appeal arguments regarding vouching on state evidentiary
rules, not on the United States Constitution or federal cases,

6 Moreover, Hoglund has shown no reason all three experts could not
have testified they saw A.H. for sexual abuse allegations. Even with no
hearsay, the jury could have heard A.H. saw professionals several times
over multiple years about the abuse. Furthermore, very likely there was
(or easily could have been) a foundation for the medical exception to But-
ler’s hearsay. But we do not rely on any of this.
32                                                          No. 18-2949

and he never argued on direct appeal that the admission of
vouching rose to the level of a federal constitutional violation.
Given that fair presentment does not require hypertechnical
congruency, that we interpret generously for pro se petition-
ers, and that the Indiana Supreme Court seemed to under-
stand the potential due process implications of the claim, we
conclude it was not procedurally defaulted.
B. Merits 7
    Hoglund invokes Howard v. O’Sullivan: “To be of constitu-
tional import, an erroneous evidentiary ruling must be so
prejudicial that it compromises the petitioner’s due process
right to a fundamentally fair trial.” 

, 723–24 (7th
Cir. 1999). This is when the error “produced a significant like-
lihood that an innocent person has been convicted.”

    No doubt the experts indirectly and directly vouched. On
direct appeal, the Supreme Court overruled Lawrence but af-
firmed because the admissions were harmless given the sub-
stantial independent evidence of guilt from A.H.’s lengthy

     7  Two Indiana appellate courts issued reasoned decisions against
Hoglund on vouching. In 2012, the direct-appeal Supreme Court ad-
dressed vouching in terms of the state evidentiary rule and due process,
and not in terms of ineffective assistance. In 2016, the post-conviction
Court of Appeals addressed vouching in terms of ineffective assistance,
and not in terms of due process. The district judge held the state courts’—
plural—determination that the error did not prejudice Hoglund was not
unreasonable. But the district judge certified the appealability of the
vouching issue only in due process terms. On vouching, Hoglund only
asks us to review the 2012 Supreme Court decision. The government by
its citations seems to think on page 16 of its response that we review the
2012 Supreme Court decision, but on page 17 that we review the 2016
Court of Appeals decision. We review the 2012 Supreme Court decision.
But either way the outcome would be the same.
No. 18-2949                                                   33

testimony, consistent and unshaken under aggressive cross
examination. (The post-conviction Court of Appeals later con-
cluded trial counsel’s failure to make consistent and thorough
objections to indirect vouching was not ineffective assistance
because at the time of trial Lawrence was still good law.)
    Appellee argues Hoglund does not offer any clearly estab-
lished Supreme Court precedent on expert vouching that his
conviction violates. His attempt to use analogous cases on
prosecutorial vouching reflects the fact that there is no clearly
established Supreme Court precedent on expert vouching in
this context. But even accepting the standard advanced by
Hoglund from Howard, we conclude he is not entitled to ha-
beas relief. The errors at trial regarding vouching were not so
prejudicial that they compromised Hoglund’s due process
right to a fundamentally fair trial. The errors did not cause a
significant likelihood an innocent person was convicted.
    Hoglund’s fundamental premise is that without the ex-
perts’ vouching (and hearsay), the case was merely “he-said,
she-said,” the case was “not a strong one,” the weight of the
evidence against Hoglund “was not great,” “the evidence is
at best in equipoise.” Hoglund points us to Jordan v. Hepp,
where we wrote that when a prosecutor improperly vouches
“and the case is not otherwise a strong one, prejudice to the
cause of the accused is so highly probable that we are not jus-
tified in assuming its nonexistence.” 

, 848 (7th Cir.
2016) (internal quotation marks omitted). Hoglund notes that
the district judge below recognized the evidence against
Hoglund (without the problems) was “not overwhelming.”
    We disagree with Hoglund. The case against him without
the problems (even broadly construing the problems) was
strong. The evidence was not in equipoise. True, the district
34                                                No. 18-2949

judge thought the case was not overwhelming. But he said
multiple times he thought it was strong even without the ex-
perts. We agree. A.H.’s testimony was strong, compelling,
graphic, often detailed, and consistent. Extensive, aggressive,
even tedious cross did not shake or expose A.H. Hoglund’s
own statements were sometimes suspicious and incriminat-
ing. The vouching did not produce a significant likelihood an
innocent person was convicted. The state court did not reach
a decision on this issue that was contrary to, or involved an
unreasonable application of, clearly established federal law.
                        VII. Conclusion
     We affirm the denial of habeas relief.