Botty Challenges

[mcgoo]

yes turnover. Had a really good run but % will reduce over larger data sets there is no doubt

[PeterLe]

Morning
Just a few thoughts...
On any bot, worthwhile looking a at the full bet history from your bet fair account (keep records even if yo make a subtle change). Extract to excel and analyse......then check to see if you had only placed the back bets or only placed the lay bets would you have been in profit? Can tell you a lot over the course of time
Also, if you do have to green up, only use the reverse setting..yes, you won't get everyone matched, but that will equal itself over time. Everytime you green at current odds, you are potentially throwing money away
Regards
Peter

[mcgoo]

Thanks Peter.Good advice. I was only looking a the reverse green up more closely today, funny enough. I have taken notes of all my changes but only in the last month & this has helped a great deal. Good to know I am on the right track.I'll start looking a the bet stats more.Cheers

[spreadbetting]

Morning
Just a few thoughts...
On any bot, worthwhile looking a at the full bet history from your bet fair account (keep records even if yo make a subtle change). Extract to excel and analyse......then check to see if you had only placed the back bets or only placed the lay bets would you have been in profit? Can tell you a lot over the course of time
Also, if you do have to green up, only use the reverse setting..yes, you won't get everyone matched, but that will equal itself over time. Everytime you green at current odds, you are potentially throwing money away
Regards
Peter

Couldn't agree more, Peter, you find most botter's, and traders to an extent, put all their efforts into the entry point and then throw away any value they've obtained with sloppy exits. Dobbers are the same, put their efforts into the entry with choice of selection, then close out with no real thought or effort behind it.

[sionascaig]

Guilty as charged - thanks for the advice Peter )

My original bot did as you suggested (just placed lay bets) and agree it a great way to get insight.

Really should go back and have a look as you suggest...

[ruthlessimon]

Something that I like tracking is my drawdown/updraw on a trade.

It's one of the few things that can boost an edge without affecting the frequency/scale.

If all my winners had a max drawdown of 8ticks, that's roughly where the stop should go. Immediately that cuts the biggest losers. In theory, targets could also be optimised in a similar way.

[ruthlessimon]

Something I'm literally looking at in my own data atm is my "losers vs winners entry price count".

In theory, a random strategy, should have an equal distribution between wins & loses, at any entry price (assuming the wins/losses are equal). What I found with this strategy - entries around mid 2s - this is not happening; & the losers are far outweighing the winners (over a range of prices - almost a normal distribution).

This is why I like just sending out hundreds of random "bots" (although I use excel vs actual trading). Use them as reconnaissance to build something even bigger

[sionascaig]

Food for thought there... Its not what I would have expected...

Looking at the drawdown/up side on the trades certainly sounds like its something worth pursuing.

I don't think what I have is even close to being optimised so work to do there. Just happy to start seeing some results come through even if it turns out to be a temporary thing )

[ShaunWhite]

This is why I like just sending out hundreds of random "bots" (although I use excel vs actual trading).

Strategy dev 101 question, what prices are you using for your entries and/or exits, best, reverse or mid? (in excel)

Using best price I find it almost impossible to find black numbers, and using reverse prices I find lots of edges but because they're slim so I can't be sure they can be executed.

I've settled on reverse for entry and best for exits but I'm especially happy with it.

[ruthlessimon]

Strategy dev 101 question, what prices are you using for your entries and/or exits, best, reverse or mid? (in excel)

Using best price I find it almost impossible to find black numbers, and using reverse prices I find lots of edges but because they're slim so I can't be sure they can be executed.

I've settled on reverse for entry and best for exits but I'm especially happy with it.

I believe it's measuring simply which back price is available. It is certainly an assumption - & the results will have an error component. The only way I can mitigate it is by assuming the targets are always hit perfectly, & the stops are always slipped (assuming the strat has stops/targets).

How would you define slim? that to me is quite subjective

For example, building a scalping strategy on my dataset, & expecting the results to be similar, would be pretty silly. However finding the markets that have the lowest volatility, using a breakout strategy (which never fires) - implies the optimal variables for scalping. Assuming we avoid curve fitting

[ShaunWhite]

How would you define slim? that to me is quite subjective

>0% and <1%

[ruthlessimon]

>0% and <1%

& out of interest how are those figures generated?

I literally work on frequency & net ticks won/lost to determine "edge slimness" - hence that subjectivity

[ruthlessimon]

Using best price I find it almost impossible to find black numbers

For example today, I've been looking at testing the break of "x" price. Here were the results, ordered via the distance to entry:

2ticks to "x" = +84
3-10ticks to "x" = +237
10-20ticks to "x" = -4
20-30ticks to "x" = 6

I actually was surprised by those figures. But a move of at least 3 ticks, I'd have thought, guarantees a market order would have filled Moreover, for a manual trader, it's easier to spot vs a 2 tick move to a breakout

To me, a slim edge is the +6 on the 20-30 variation

[ShaunWhite]

>0% and <1%

& out of interest how are those figures generated?

I always use simulated trading with greening and commission, I don't count ticks because they're all worth different amounts at different places. Ditto the effect of greening, depending what direction you're going.

[ShaunWhite]

2ticks to "x" = +84

2 ticks is noise not a distance ☺

I do all this analysis and come to a conclusion that just a tick or two per market most of the time and no big losses is all that's needed (iykwim). Then i think wtf am I doing it excel, I can find that looking at the ladder.