Sunday, 22 April 2012

Enough Value

Peter Webb wrote a long post on the Lay The Draw. Big Al called it ‘an interesting post’, although to me, it added nothing substantial to the debate, and I was planning to ignore it, but after consideration, I’ll make a few comments.

No one is suggesting that a strategy as simple as laying every draw or backing every draw can ever work, so looking at 15,000 matches is of limited value. Peter also suggests that you can have value, but then differentiates between having value, and having ‘enough’ value. Value is value. If you stake using Kelly, or more realistically a fractional Kelly, the profits will come. Your stakes need to be appropriate for your selections, and for your ability to tolerate losses. The ability to handle a losing run, and no system can ever guarantee short losing runs, is a requirement for successful trading, but if you are finding value and can’t handle a losing run, then what exactly is the ‘enough value' that WILL allow you to handle losing runs?

I’m not sure what period the 15,000 matches Peter looked at spanned, but markets do change. As I mentioned a few days ago, even Ian Erskine, chief Prophet of the Lay The Draw philosophy, has recognized this, and is planning a tweak, although why he feels the markets have left the door open and vulnerable to this change is unclear. I have no idea how many apostles Ian has, but it’s reasonable to think that their involvement in the Match Odds markets for the recommended matches does have some effect on prices both pre-game and in-play.

As Peter says, edges do indeed fade to nothing. With the subjective element in the Lay The Draw system, it’s very hard to say definitively that it works or it doesn’t work. Some will hold their position after the first goal, (assuming there IS a first goal), others will trade out immediately, while others will wait for a while before trading out – so all traders will experience different outcomes.

XX Draw subscribers might well be trading the games, and laying off after 80 minutes or whatever if the game is drawn, but by simplifying the record keeping, it does make it easy to see the value (or not) of these selections.

It’s for this reason that I include Ian’s FTS selections in the table. Selections are exposed to the ‘world of truth’ I mentioned yesterday. A short sample like this (78) proves nothing of course, but then nor does a sample of 15,000 random matches. GIGO is the term I recall from my early years in IT.

It can be a fine line between using a large enough sample to validate your edge, and too large a time period so that your 'edge' evaporates.

Here is Peter’s post in full, or you can read it here:

Whenever you look at all strategies, no matter how complex, there is always some simple underlying key metric that is essential to whether a strategy works or not. This is also true in the case of ‘lay the draw’.

I’ve recently been looking again at value backing. There is a little more depth behind what I am doing, but one of the key issues you face if you pursue a value strategy is deviation from mean. You may be able to identify value, but if you can’t identify enough, you run the risk that an unfavourable run of results will see you draw down a significant amount on your capital before your judgements prove correct. Most people can’t bear this unless they are very patient and have very deep pockets. Neither tend to be realistic of any but the most experienced.

One of the areas I have been looking in-depth is at the moment, is the draw on football matches. I didn’t look at a few days, weeks or months but at over 15,000 matches. Below you can see a graph of the data I collected. In theory, if the market was efficient, you wouldn’t see much variation from the centre line. However, this only tends to occur over very long time periods. You would think that 15,000 matches would be defined as a long period. But from the graph however you can see that there has been a significant variation in returns from the draw over the sample period.

Basically the graph is showing you what would happen if you backed the draw. Any variation to the downside of this graph would be a net positive variance for people who lay the draw and the opposite is true. Of course this would also apply to people trading as well. If a match ends as a draw or not, there is a demonstrable chance you would have traded it correctly dependent on that result. The blue line is what happened to level stakes. In this case using £10 you would have faced a drawdown of £4.7k, the other line is my attempt at seeking value. This drew down as well but eventually recovered to break even.

The upshot of this data is even if you had your value judgement completely wrong you could have been fooled into thinking you had stumbled on some amazing new system, or that you were the god of tipsters. But the simple fact underlying your strategy would have been that you laid rather than backed. Essentially you could have ended up over a long period of time if you had the faith to stick to your initial thought. The problem of course is that, even then, your ‘edge’ would have eventually faded to nothing.

This isn’t just applicable to the draw, most events and selections in those events can go through these huge under or over runs. This tends to lead people to the view that somehow they have stumbled on something when in fact the market is just conforming to its natural variation. Anybody can be a ‘god’ for a short period of time, or in this case over 15,000 matches. But doing that over many, many years is much harder. When looking for advice or somebody to look up to, longevity is a key differentiator.
I'll tidy up the formatting later - Blogger sucks!

1 comment:

BigAl said...

I thought it was quite interesting in that a blind man with no idea about football betting could have made decent profits backing the draw in recent times.

Ho hum.