@Adam said in John’s Horse Trading Resources:

@Tony-Hastie Respect! I only got into running in the last few years and am fairly comfortable with a half marathon. Ultras still seem like a complete pipedream though!

Look at you too. I’m just happy with my 5 to 10 k runs. My daughter made me feel old yesterday. She ran 10k in under 50 minutes. I average 35 minutes for 5k 😂

@Adam Fantastic Thanks for the Update

@Tony-Hastie It's an impact value figure that tells you how much more or less winners each draw position produces, compared to what could be considered its fair share.

A value of 1.25 means there are 25% more winners than you might expect.

So for example if a course hypothetically had 4 stalls, you might expect an even 25% of wins in each stall if there was no bias. If one of the stalls gets more than 25% of the wins, its draw bias figure would be > 1.0.

It's narrowed down based on course, race type, and distance.

@Adam has done amazing work on the Jockey and Trainer stats! So many options now to help make a good strategy

@Paul-Hill don't think we could even investigate that, as there is no strategy to work from. perhaps something in the betting rules was slightly different, impossible to know if it wasn't saved unfortunately.

if you are 100% sure create it again and see if the results align with the saved one, save both and let us know the names if there is an issue and we could investigate.

@Paul-Hill The outsider is the horse with the highest price.

2nd outsider is the horse with the 2nd highest price, etc. Basically the opposite of the favourites.

The outsider setting works in conjunction with the other betting rules, so for example if you set your entry time as 10 mins before the race, it's basically going to give you whichever horse had the highest price at 10 mins before the race.

@Patrick-Murphy Having looked at some of these, it looks like they're due to gaps in the prices. So basically short amounts of time where we weren't able to capture price snapshots, so your exits have either been missed completely or recorded later than they should have been.

These gaps can be due to latency or brief outages of the Betfair API and occasionally due to some issue our end, like the server issue I posted about recently here.

We regularly plug these gaps where we can using the Betfair historical data, and we've recently improved our audit scripts to better identify when they happen, so have been going through the whole database again to apply this improved method. It's a fairly big task but should be finished by the end of next week.

Sometimes when the issue is at Betfair's end, the data doesn't even appear in their historical data, which means we won't be able to plug the gaps at all. This will represent a fraction of a percent of trades, but I can understand that if you're backing at very high odds and large stakes, it could skew the profit figure.

Maybe we need a way of marking these in the database. We could also add some kind of "report and exclude" button to each individual trade in the results table, so that if you spot any of these yourselves, you can submit them for review and make them not appear in your backtesting results.

I'll finish the DB audit and then see what we can do to deal with any issues that still remain.

@Danny-Bridges Actually I was wrong about it being a backwards compatibility issue.

It's actually being caused by a bug when you include "draw bias" in the individual trades results columns.

I'll fix this today but for now, just exclude that column in your results 👍

@Danny-Beal thanks, I have betangel, but if it is possible on bfbm I have a hope that I can somehow do it with betangel, thanks.

@Martin I'm really looking for a standard approach to calculating my stakes as a percentage of bank across various strategies (rather than based on a specific strategy). Generally I'm backing but some lay strats.
So for example, one strategy might have a 4% strike rate, odds range 20-100, average odds 38, longest losing run 133 and an ROI of 24%. I know I'll need to ride out long losing periods and my gut feel is I'd want at least a 200 point bank, probably more. If I plug the above into a basic Kelly calculator it gives me 1.41% as the 'Fraction of bankroll to wager' which equals a 70 point bank. That's way too low in my view.
2nd scenario - 24% strike rate, average odds 8, longest losing run 22 and an ROI of 37%. The basic Kelly calculator gives 13.14% as my stake %age (!) but my feel is this would work with a 100 point bank.
I've seen Kelly calculators with options to add multiples for a more cautious bank size (BFBM has a 'Divisor') and maybe I just need to use that but adjust the Divisor based on the average odds?
I've always used a level stakes approach and that's probably why I'm struggling with the percentage of bank that Kelly calculates. I also need to adjust to the percent of bank betting approach rather level stakes
Anyway, any thoughts from others with more experience using a percentage approach will be very welcome
Cheers

@nigel-stone welcome it's great to have you here! if you get stuck just let me know

@Andrew-Hopkins For today, yes

That's massive bad luck, for two horses at single figure odds to be withdrawn after the off time and then such a big priced outsider going on to win before a retrospective rule four is enforced. Going to be a very rare occurrence, but can be pretty painful when it does happen.

That's why I tend to do this type of trade with National Hunt races only because there's no messing about with stalls and you can see exactly how everyone is lining up.

@Martin thank you as usual Martin

@Danny-Newby It will be horses that are expected to be held up. If they don't yet have enough history to satisfy the pace calculation, their rating will be null and the horse won't appear in the results.

Nice one Adam.

Yea I would be more of the thinking that if certain rules should not be selected at the same time it should just auto unselect them (maybe with a brief flashing animation of the unselected option, to alert user to what just happened)

Then it will be obvious that those two are incompatible (You could do this with the Max Qualifiers and Fav settings as well, same logic as we previously discussed, isn't it)

Cheers

@simon-topper said in The Forward Testing Thread:

@John-Folan This one looks interesting, set it up myself as well.

Bad downturn in September but only 10% of the years profit

@Andrew Ok, thanks

Has anyone else found that quite a few strategies seem to have almost flipped round when you compare 2022 with 2023 (nearly a full year of results). Guess the challenge is to crack that and find something that is consistent over the two years of data (I'm ignoring 2021 for now due to covid disruption). Any thoughts?