Trading What I see !?

[ShaunWhite]

Thanks Shaun, yes im actually running some analysis on better staking based on your comments yesterday on how you would have scaled back.. I m dubious on it, but we'll see what some testing tells me....

I don't vary stake in tests because there's no firm logic to when to increase or reduce. I just use it to keep me sane, eg based on the the last few days/week how confident am I feeling. That's much like manual trading I guess where you look at the situation and stake accordingly.. Some phases feel like a 'Cheltenham' with little variation and you can go hard but other spells might look a bit wild and I'll play safe. I'm not sure it could turn a loser into a winner but it seems a bit nuts to just steam ahead on max stake.

[ShaunWhite]

Now you've got a good bet history try having a look at the time your offers take to be matched. (eg sort your cumulative PL by that). You might find some sort of fill/kill will help, stale orders can be a weak point.

[goat68]

Just run a backtest of the greyhound bot on my data from Sep'21->Jan'22 just to make sure it repeats what i'm seeing live, and it's quite a striking change in trend:

Screenshot 2022-02-02 at 20.17.33.png

[goat68]

Ran a modified version using variable staking being risk averse when the trend is down.. which is better?

Screenshot 2022-02-02 at 21.17.18.png

[ShaunWhite]

Ran a modified version using variable staking being risk averse when the trend is down.. which is better?
Screenshot 2022-02-02 at 21.17.18.png

The one that makes most money is always best. I've no idea how you worked out your variable staking

[ShaunWhite]

Just run a backtest of the greyhound bot on my data from Sep'21->Jan'22 just to make sure it repeats what i'm seeing live, and it's quite a striking change in trend:

It's a change but not that unusual, I'll try to dig one out I had that was the same but then turned back to normal just as fast. Randomness does strange things and other people come and go. History can only tell you so much so the decision to go for it or not is always subjective.


...found it.
Lasted for 4 days, about 2,500 selections. Came back but was never quite as good. Any explaination is pure speculation.

Screenshot_9.jpg

[goat68]

Ran a modified version using variable staking being risk averse when the trend is down.. which is better?
Screenshot 2022-02-02 at 21.17.18.png

The one that makes most money is always best. I've no idea how you worked out your variable staking

Staking was simple, reduce stake when PnL trends down, increase when trends back up !

[goat68]

Just run a backtest of the greyhound bot on my data from Sep'21->Jan'22 just to make sure it repeats what i'm seeing live, and it's quite a striking change in trend:

It's a change but not that unusual, I'll try to dig one out I had that was the same but then turned back to normal just as fast. Randomness does strange things and other people come and go. History can only tell you so much so the decision to go for it or not is always subjective.


...found it.
Lasted for 4 days, about 2,500 selections. Came back but was never quite as good. Any explaination is pure speculation. Screenshot_9.jpg

This game is basically about taking other peoples money! So it might depend who was taking my offers, I can't believe that would be just 1 or 2 opponents, and thus if those 1 or 2 changed my strategy would suddenly disappear, more like a general market behaviour which would disappear more slowly if it did...
Thus i'm hoping it will come back... it's been 2 weeks so far...

[ShaunWhite]

[Staking was simple, reduce stake when PnL trends down, increase when trends back up !

You make that sound simple but I think you know what I mean, what's a trend? what's an increase? are reductions and increases at the same rate? what's the upper/lower bounds of the stake etc etc. Again you seem to do something in an evening that should take days or weeks.

[ShaunWhite]

This game is basically about taking other peoples money! So it might depend who was taking my offers, I can't believe that would be just 1 or 2 opponents, and thus if those 1 or 2 changed my strategy would suddenly disappear, more like a general market behaviour which would disappear more slowly if it did...
Thus i'm hoping it will come back... it's been 2 weeks so far...

Have you looked for anything concrete? Has your fill rate or time to fill changed? Has your bet latency been different, And if so are your offers getting stale and a fill/kill could help. Speciation about other people is a waste of time, it's an known unknown as Donald Runsfeld would say.

[CallumPerry]

i'm seeing some similarities in our strategies Simon. I believe this is November-November for me.

Is there a problem with it Cal?

So long as it hasn't been super overfitted it's worth a punt imo. Just keep an eye on the max drawdown & mentally commit £750-500ish as lost (or drop the stake)

Yes, turn yours off now

If it is the same/similar then I'd bet good money it's not just me and you doing it, absolutely no problem in that. It's working for both of us. My maximum down was about £400 so I'm happy to leave it for another 12 months to observe and hopefully grow.

[CallumPerry]

You should feel great about being around the breakeven stage Goat. Like Shaun said, stick to your guns, greyhounds auto seems a good area for you (much like myself), continue to heavily analyse in creative ways. Approach the problem(s) from many angles and an ever so slight % ^^^ will result in a constant little continuous stream of passive income.

I'm still only at the little stream part myself but making enough to cover a 24/7 server, keep BF sweet with collecting even more data and building my funds for bigger ideas is a great feeling. You'll be there soon enough and probably surpass me easily Keep going!

[goat68]

You should feel great about being around the breakeven stage Goat. Like Shaun said, stick to your guns, greyhounds auto seems a good area for you (much like myself), continue to heavily analyse in creative ways. Approach the problem(s) from many angles and an ever so slight % ^^^ will result in a constant little continuous stream of passive income.

I'm still only at the little stream part myself but making enough to cover a 24/7 server, keep BF sweet with collecting even more data and building my funds for bigger ideas is a great feeling. You'll be there soon enough and probably surpass me easily Keep going!

Thanks Callum, and hope your bots are doing well also.
Cheers
G

[goat68]

So i've done some analysis last night, ran some test during today and made some improvements, the main one being my bet fill&kill logic, which has actually boosted my profit and also flattened the recent drawdown quite a bit.
I also evaluated the variable stake risk-averse strategy, and like just about any stop-loss type of thing it reduces losses, but reduces overall profit. So thinking about it in a 5month backtest sample it reduced the max drawdown by £100+ or so, but the overall profit was £400 lower!, so not worth it, better to take the ups & downs and just keep an eye on things....

Anyway, here's the previous graph with the improvements:

Screenshot 2022-02-03 at 22.46.37.png

[goat68]

So done a "NewImproved_2" , which is actually a slight simplification, but amazingly it shows no drawdown during the january poor period. Great, but when is it going to really come to "live" fruition?!

Screenshot 2022-02-05 at 08.06.49.png