Required Strike Rate Formula

[thomsch]

I'd rather know if the strategy was +ve expectancy in the 1st place; rather than what strikerate I need to make it >0

I take it you would use back testing to know whether a strategy was +ve expectancy in the first place? Do you know how to do this for football?

I can see the logic in that; but the bit that troubles me, is over what period the strikerate gets measured.

I'm really not sure, probably 1000s of trades and is probably a major flaw in this metric? I think I will find it useful though as a quick guide to tell me how well I'm performing when reading a football match for a goal. It may also boost my confidence in a strategy to keep using it if the ASR is a lot higher than the RST.

2. Expected expectancy Basically the idea being, we assume the expectancy of strategy remains constant, & plot the equity curve into the future. We then add 2 boundaries to this line (upper & lower i.e. +/-5%); & if the real p&l breaks the lower bound - the strategy gets turned off. I think that's faster than max drawdown; but I haven't done it yet

This is interesting and something to look in to. I am by no means an expert with Excel, would it be easy enough to do?

[ruthlessimon]

I take it you would use back testing to know whether a strategy was +ve expectancy in the first place? Do you know how to do this for football?

A quick (cold) example.

I've often wondered whether the Super6 "what people are saying" is connected to the pricing of the match.

For example, if 94% of participants think Chelsea vs. Everton, is a win for Chelsea - is it profitable to back teams (@ any price, @kickoff), if the "what people are saying" is >90%?

So, I'm gonna go out collecting the data, splitting it into in/out sample, optimise the variables (i.e. >90%), test it on the out-of-sample. & if the profitability is "similar" on the out of sample, it gets moved onto sim for a bit, still good, then live small.

Most stuff fails, when it reaches the out-of-sample data. I have very few things that genuinely pass that stage

Also, if someone could explain, how that idea above, gets turned into a "model" I'm all ears

This is interesting and something to look in to. I am by no means an expert with Excel, would it be easy enough to do?

I haven't done it myself yet; but I like the logic - & I'm pretty sure (done right), is better than any drawdown metric

[ShaunWhite]

This is interesting and something to look in to. I am by no means an expert with Excel, would it be easy enough to do?

I haven't done it myself yet; but I like the logic - & I'm pretty sure (done right), is better than any drawdown metric

See above. TREND and STEYX
Extrapolate the high and low and then try to stay betwen them when you go from test to live. As I said above StdErr is a bit tight so you might want to add a factor to it to give you some wiggle room.

Or are you describing something different?

[ruthlessimon]

See above. TREND and STEYX
Extrapolate the high and low and then try to stay betwen them when you go from test to live.

Ty, looks legnd

[ruthlessimon]

StdErr is a bit tight imo so you might want to tweak it a bit.

Problem is, that then becomes a variable

What's the difference between optimisng that, & optimisng an entry criteria?

[ShaunWhite]

StdErr is a bit tight imo so you might want to tweak it a bit.

Problem is, that then becomes a variable

A person's pain threshold is a usually contsant not a valiable. Set and forget.
StdErr wasn't designed for trading but it's a fair start given that it accounts for volatility. I guess your +/- 5% idea came from eyeballing things and (shock horror) 'feeling' what looked right....this isn't much different but adds a qualative figure and not just a hunch

[CallumPerry]

Don't want to get all soppy but I have missed you Shaun. The forum hasn't been the same without your grumpy old arse and you do say some very sensible things often which helps me add another level to my system.

[thomsch]

Thanks for the trend analysis spreadsheet ShaunWhite!

[brimson25]

Thanks for the trend analysis spreadsheet ShaunWhite!

Second this, just what I wanted. Thank you.

[Dublin_Flyer]

A quick (cold) example.

I've often wondered whether the Super6 "what people are saying" is connected to the pricing of the match.

For example, if 94% of participants think Chelsea vs. Everton, is a win for Chelsea - is it profitable to back teams (@ any price, @kickoff), if the "what people are saying" is >90%?

I tried something similar using free source data, I think it was the Livescore or Forza app where you can predict who will win a given match. Using 1000 votes as a minimum for a match, then converting the percentage to a price to see if there was a people "edge" I couldn't make any progress out of it because not enough people ever predict a draw to come anywhere near a true reflection or betfair odds at kick off.

I think most of "what people are saying" is familiarity bias. Ask people on the street who'll win the game between Real Sociedad and Barcelo...BARCELONA. Who'll win between AC Milan and G...AC Milan!

Then if you say Ok, you said AC Milan will win, give me your £50/€50/$50 and we'll bet it on them........oh wait, you never said there was my money at risk!! What team are they playing again??

[firlandsfarm]

Would anyone be able to help me figure out the formula to calculate the required strike rate for a trading strategy i.e. what strike rate would be required to break even for a given strategy? There's plenty on the web for straight betting strategies but I cannot find much for trading.

As an example, for an average win of £5 and average loss of £2 a strike rate of 28.6 works out about break even. I did this through trial and error but could not figure out the formula to calculate it quickly. Does anyone have any ideas?

Only just spotted this thread because it was resurrected by Dublin-Flyer's post earlier today and although there has been a lot of theory put forward and complicated calculations suggested in answer to your initial question I just thought I would offer the simple solution!

Calculation
Average win £5
Average loss £2
Look at it as an odd = 5/2 and allow for commission say at 2% but substitute your own commission rate
Decimal is 5/2*0.98 + 1 = 3.45
100 races, Win's required = 100/3.45 = 28.98
Therefore losing races = 100 - 28.98 = 71.02

Proof
Winning returns = 28.98 * £5 * 98% = £142.00
Losses = 71.02 * £2 = £142.04

Simple really!

[Dublin_Flyer]

Brimson resurrected it not me, I just noticed that there was a 3 hour gap between his post and mine, then 18 months between his and the previous one! The things that happen when you click on active topics

[firlandsfarm]

Brimson resurrected it not me, I just noticed that there was a 3 hour gap between his post and mine, then 18 months between his and the previous one! The things that happen when you click on active topics

Yes, I see that now … my apologies