Trading What I see !?
September drawdown continuing... already wiped last month's profit! however in theory since I have 3 strategies in play here drawdowns of this magnitude are expected... still holding confidence in them, as they all backtest with P-values over a 4 month period of 0.02%, 1.89% and 2.2%
You do not have the required permissions to view the files attached to this post.
With drawdowns there's definitely a lot more anxiety when it's your own money, as opposed to profit money. Eg.if i had made say £1000 and was now suffering a £200 drawdown, it's easier to handle, as opposed to my current situation of a -£200 drawdown and -£560 trading lifetime....
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
"profit money" is your money.
The language of gambling is designed to make the customer less attached to their "winnings" than the other money they may have earned so like you, they care a little less about giving it back. It's a corrosive mentality you'll need to get rid of otherwise you'll always just give back your hard earned profit.
Getting a bit hacked off, these strategies are the best I can come up with and months worths of effort...
but that's what drawdowns do to you, they install doubt !You do not have the required permissions to view the files attached to this post.
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
It was only about 3 months ago when you started collecting data and learning how to test properly wasn't it? It's a frustrating game sometimes but try to remember most of your first year's stuff was using totally different methods and I (and others) did say you'd need 3 or 4 months data as a minimum before you can even start to look for anything. These aren't the best you can do just the best you've been able to do so far.
If you're getting different live results to your backtest then there must be a flaw in your methodolog or certainly might be. Until you iron that out then take these results with a large pinch of salt.
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
I'm assuming you mean the likelihood that it's just random? Those still don't rule out the spector of backfitting, in fact backfitting inherently produces low probabilities. I'm more inclined to produce a range of backtest results from multiple random samples and compare them to completely eliminate that possibility, as I explained quite a while back in this thread.
P-values are only as good as the critical thinking that went into producing them. Stated as stand alone facts they don't mean much and with all 3 being sub 3% then I'd be more sure they were wrong than being sure they were right. The one that's 0.02% is so far away from logical given your sample size that alone would make me highly suspicious, can it really be a 5000/1 on bet it's a winner?
I don't think it's that you can't find anything more a case that you're so keen to find something you've stopped being as rigorous as you need to be?
not sure that's what Pvalue quite is, it's the odds of it not achieving a 0% yield over the given sample size I think. So a probability of 0.02% that in say 5000 bets it does not achieve a 0% yield, ie.breakeven.ShaunWhite wrote: ↑Wed Sep 15, 2021 4:31 pm
P-values are only as good as the critical thinking that went into producing them. Stated as stand alone facts they don't mean much and with all 3 being sub 3% then I'd be more sure they were wrong than being sure they were right. The one that's 0.02% is so far away from logical given your sample size that alone would make me highly suspicious, can it really be a 5000/1 on bet it's a winner?
- ruthlessimon
- Posts: 2094
- Joined: Wed Mar 23, 2016 3:54 pm
So I only have 4months of data. I've run it across the whole data, and plotted, I can see how it varies, and can see the current drawdown is most likely normal. Doesn't stop me from being hacked off though
- The Silk Run
- Posts: 927
- Joined: Mon May 14, 2018 12:53 am
- Location: United Kingdom
Hi Hi G. Have you not considered a more simple approach to your stratergies. I do think your over complicating things, somewhat.
You may be pleasantly surprised
You may be pleasantly surprised
Thanks SR, my strategies are actually quite simple value bets, ie.a straight bet at a certain price, at a certain time, based on the theory Betfair prices are 0ev if you add them ALL up, but that can mean half of them are +1ev, and the other half -1ev . So i just bet on that +1ev half, simple!The Silk Run wrote: ↑Wed Sep 15, 2021 5:01 pmHi Hi G. Have you not considered a more simple approach to your stratergies. I do think your over complicating things, somewhat.
You may be pleasantly surprised
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
yes and no, p-value is the probability of obtaining results at least as extreme as the observed results of a statistical hypothesis test, assuming that the null hypothesis is correct. It's akin to the standard deviation from the null hypothesis, ie breakeven. But it's only meaningful using subsets of your data and comparing them.
If you're still using your whole dataset, and I think you said you are, then there's no way to prove you're not just backfitting. At least try using subsets and see how they compare?
You do not have the required permissions to view the files attached to this post.
Yep, I wouldn't disagree with all of this...ShaunWhite wrote: ↑Wed Sep 15, 2021 5:46 pmyes and no, p-value is the probability of obtaining results at least as extreme as the observed results of a statistical hypothesis test, assuming that the null hypothesis is correct. It's akin to the standard deviation from the null hypothesis, ie breakeven. But it's only meaningful using subsets of your data and comparing them.
Screenshot_23.jpg
If you're still using your whole dataset, and I think you said you are, then there's no way to prove you're not just backfitting. At least try using subsets and see how they compare?