Risk..
Also ive been analysing liquidity, and have some logic in the bot...
However could be in the valley of the stupid !!
It's Tuesday and could be facing a -£30 loss in the face tonight, egg on...
Risk..
ShaunWhite wrote: ↑Tue Apr 27, 2021 1:56 amWatch out for the Dunning Kruger effect. You'll be looking back on these recent posts with similar feelings in a year's time.
Screenshot_5.png
Even if you got away with it, running untested code in live is a cardinal sin. I'm afraid we'll have to escort you out of the buillding immediately and a colleague will clear your desk and leave your belongings in reception.goat68 wrote: ↑Tue Apr 27, 2021 10:45 pmSo couldn't understand why bot only entered 1 trade today, backtest on today's data showed 10 and net profit...
A bit of investigation... and a very minor change I did broke my conditions. I didn't realise that "Clearing a Signal", is not the same as setting it to 0, so my condition "if signal value < 5" didn't trigger when it was cleared (thinking it would be 0 !), DOH!
Anyway fixed... onwards and upwards...!
Haha! I prefer to run on live min stakes, dont trust practice mode, paranoid betfair will close me down!ShaunWhite wrote: ↑Tue Apr 27, 2021 11:01 pmEven if you got away with it, running untested code in live is a cardinal sin. I'm afraid we'll have to escort you out of the buillding immediately and a colleague will clear your desk and leave your belongings in reception.goat68 wrote: ↑Tue Apr 27, 2021 10:45 pmSo couldn't understand why bot only entered 1 trade today, backtest on today's data showed 10 and net profit...
A bit of investigation... and a very minor change I did broke my conditions. I didn't realise that "Clearing a Signal", is not the same as setting it to 0, so my condition "if signal value < 5" didn't trigger when it was cleared (thinking it would be 0 !), DOH!
Anyway fixed... onwards and upwards...!![]()
When you posted that did you think 222 pages later it would still be going strong?
is the consensus dont increase stake until r2 nearer 1 then?
Hi, so I think for trend analysis R2 of >0.95 is the typical mark of a good trend, something near that would satisfy me.michaelhud1 wrote: ↑Wed Apr 28, 2021 9:06 amis the consensus dont increase stake until r2 nearer 1 then?
i believe ive come across my first profitable strategy projected £60 net profit from 1000 markets but r2 very low, i was getting excited but after comparing with your results (much better than mine) maybe i am clser to mount stupid
(edit:i undertsand the closer to 1 to more confidence but my brain see over .75 as confirmation?)
I suggested using r-squared purely as a guide. And what's an acceptable value depends on how the strategy behaves. A hgh strike rate would show a high correlation to trend and something with a low strike rate would'nt be correleated so closely. I definately wouldn't use that on it's own to make any decisions.
Yep PM is purely to test it functions correctly, not to see if it's profitable or not. And sub £2 in-running bets are going to take twice as long to hit the market anyway so won't mean much.firlandsfarm wrote: ↑Wed Apr 28, 2021 8:31 amBTW I share your 'distrust' for the practice mode, as some constantly remind us practice mode can never fully reflect the live market because once you switch from practice to live you change the live market! In-Running is especially sensitive ... I will test new code in the live market for below minimum stakes, maybe just 10p for the first run then up it to £1 when all seems OK and finally go for live stakes.
I don't know how big your shop is but by the time you start to affect a pre-race horse market with one fairly long duration bet you'll probably be staking in the thouands. HFT trading a pretty size sensitive but your method of using one bet to open and one bet to get out, several minutes apart, are much much less sensitve. Be careful not to just oscillate between steps 1 and 3 on your 'journey'.goat68 wrote: ↑Wed Apr 28, 2021 10:39 ami'm increasing stakes not on the full expectancy that this bot is profitable, but as a medium-term test of various "scaling" alterations I have done to the bot, including "liquidity evaluation", "over staking check", all based on my own thoughts and opinion, hence needs a bit of a live test. If it costs me half the price of my weekly supermarket shop then i'm happy with that !!