There are good gains from octane since you can run a lot more timing.
This is my 02 Forester. The Yellow line is current mods with an I-Speed SRS-10 flash (87 octane). The purple line is the torque curve with an I-Speed SRS-20 flash (91 octane). There were zero changes to the engines and the data logging was done on the same day one right after the other. The results were middle of the road results of I believe 6 pulls from 1500 to redline for each set. For reference, the original dyno was my car with a stock engine and only a SRS-10 flash, otherwise completely stock, done on a Dyno Dynamics dyno. All my data logging since has been scaled to that original real dyno pull. Results are pretty accurate up to 4500rpm, but then road logging lags behind some in results with it being low by 15hp/ft-lbs by 6500 versus a real dyno. This had been figured at a later date with another real dyno pull on a Mustang dyno and compared directly with road dyno results of the same setup.
The raw numbers are no important as they are dependent on the scaling. The differences however are pretty accurate within 1-2hp/ft-lbs and all results were done with non-favoring results (high or low) taken over at least 3 runs but more often 4 to 6 runs on a specific road each way in order to verify wind/inclination were not factors or at least could be adjusted not to matter since I have 8, 10, 12 results to work through and pick the best average representation.
It might be curious to try higher octane. However, you only have so much timing to play with before it kind of gets pointless. Then, you're pretty much forced to someone generate higher dynamic compression (pistons, variable valve timing) to make actual use of the knock resistance.
For reference, timing on the SRS-10 flash (87 octane) versus the SRS-20 flash (91 octane):
I think both gasoline and E85 (if you so choose) can show gains upwards of around 40 degrees of advance, although don't quote me because frankly I've researched this subject matter about zero. I just know that above a certain point, you're not gaining, and that gasoline and E85 is pretty similar in this trait, at least from the brief amount I've attempted to search and read. I'm not a tuner and don't deal with this in any way, so take my comment with a pile of salt and spend time researching the matter yourself. I can only compare a 87 octane setup versus a 91 octane setup, show power gains, and you can guess and hopefully research the rest.
p.s. TWE headers are the only ones you should be looking at if you're actually serious about power gains. If not, you're only fighting your own goals.