) Interested to know what people make of it. To me, it looks "too good to be true" so I'd welcome some more professional eyes/brains to help me qualify this or to help me dig deeper.
There is a lot of stuff that went into this, I will not bore everyone with ALL the details (unless anyone is interested)
Some relevant details though, first:
- This is a completely automated system, based on data collected from my bespoke web app (so i believe it is probably unique data in the precise windows I am collecting it)
- It involves betting on literally every single horse, either Backing or Laying
- Results are not to be taken too literally, as they are based on £50 back bets, and lay bets of a more complex nature but ranging in £50-250 liability depending on odds. It might be that dumping £50 on an outsider may of course drop the BSP odds somewhat in reality, but for argument sake, you could knock a zero off all these numbers and assume £5 bets, it should probably be accurate for that to within 1% I would guess.
The Numbers:
- The top table reflects individual bet stats
- The bottom table reflects total race stats, but also shows a bigger total profit, this is because it also includes 'accurate commission' over the whole race, whereas the top table assumes 2% paid on all winners, regardless of reductions for losing bets on the same race
- The data spans every horse from mid march last year until the end of January (I've not finished adding Feb/March data to the database properly yet but i know it's not changed things too dramatically)
Now, seeing as some of these wins come from backing huge price outsiders that win, you could be forgiven for thinking that maybe this is probably just short term variance showing a positive result. But over this many horses, 106K bets in total, and commission included - what do you think the chances are that I have found something good here? Obviously it's hard to say for sure without me revealing the specifics of the system, but Grok told me that over this much data the p-value would essentially mean the chances this is down to luck is virtually impossible. Looking at the chart, it looks about as good as one could expect, doesn't it?? Or am I missing something?
I started live testing it at the start of the year with £1 bets, balance went generally slowly down by a few hundred over the first couple of months, but in the first week of April alone, it's gone up by £800 putting me in a good net profit for the year, and lining up nicely for what Grok told me: Generally systems that perform well over summer will NOT do well over winter, so expect an upturn in results again starting from the end of March/early April.
I would really appreciate any constructive questions I could be asking to figure out whether this data is looking reliable or if somehow the results could be too cherry-picked, even over data this size. The max back odds I am using for this particular set is 200 but it doesn't change much regardless of changing that to 100 or 1000. I'm not sure if there's enough data to consider the 3-digit odds data as being reliable, but I think you could remove my 10x biggest winners here and it would still be in a very nice profit.. and i'd take that!
