Black Cats and T-Bones

Model Trader had a comment on my Top Notch post, writing:

I'm interested that you warn against waiting for statistical proof before getting involved? I can see why you might not want to wait for the routinely accepted definition of 'proof', where for certain areas, the bar is very high, but surely calculating some statistics (what is the likelihood that the edge occurred by chance) can only help?

I imagine that many times it's common sense. If I flip a coin 100 times and get 90 heads, I'm confident the coin is biased. If I get 51 heads, I intuitively know that I probably don't have an edge. But what about 55, 60, 65? Where do I draw the line? And let's say that I believe the 65 out of 100 coin is biased with a 0.65 probability of heads. Once I've started betting there's a very good chance that even with that edge, I'll be down at some point, so how do I decide when to get out because the edge has gone, rather than because of random chance?  

Model Trader ask some good questions. My thoughts on this are that betting markets are not ideally suited to the traditional standards of proof, mainly because they are constantly evolving and adapting and if any edge is present, it should soon be discovered and the market will adapt accordingly. That's why I suggest that if you wait for statistical proof, you'll have missed the boat, and the edge will have sailed.

Statistical proof is great for something like controlled medical studies, but essentially useless for sports betting where the environment is constantly changing. Rules change, players change, strategies change. Does anyone really expect to study results from the ATP or WTA and be able to have statistical confidence in a system that has been profitable in the past? I would hope not.

I think where the idea of statistical proof does have some merit in betting is when evaluating a tipster's record. Joseph Buchdahl wrote a book on this called "How To Find A Black Cat In a Coal Cellar". Well worth a read, but for those with shorter attention spans, or retired NFL players, this shorter article is also very interesting. It concludes:

However, until I see some significant sample of data that contradicts everything I have reported on here, I am prepared to stand by my assertion that almost all so-called tipsters are anything but, and that most of what is being sold represents nothing more than slick marketing in an attempt to attract customers to products that in fact are not what they claim to be. For the most part, that is not because the sellers are trying to defraud the buyers, but simply because the sellers haven't yet realised that luck is not the same thing as skill.

How to decide when the edge has gone is a tricky question. Where to draw the line is a personal decision. For me, it depends on the confidence I had in the strategy in the first place, and I suggested a few days back using a stop-loss while continuing to track results. Monitoring the ROI can be useful too, although since most sports have a relatively small number of bets in a season, I wouldn't put too much weight on a percentage based on less than 100 bets.

One example about confidence is the T-Bone System, which I've written about several times, and which has three components - Money Line, Run Line and Overs. Last season, in July, I wrote a post about the large profits from backing the Overs in these games. I wrote at the time:

Whilst both the Money Line and Run Line profits are steady enough in the last five seasons, the profitability of the Overs component is something new this season, and rather a mystery.

So I have far more confidence in the Money Line (profitable in 6 of the last 7 seasons) and Run Line (profitable six seasons in a row) part of this system than I do in the Overs which is basically flat over seven seasons. 

Hopefully some of you are playing this simple system this season. 

Eggmund returned with a comment on my Wall Street Journal Best Stock Award post from yesterday commenting:

Looks like an award for building a volatile portfolio, not one that will necessarily perform in the long run. Not a particularly meaningful award, apart from telling you which managers will give you the most exciting roller-coaster ride!

I wouldn't expect an index tracker to win this award even if eligible as they should track the market rather than massively out / under-perform it. That's not to criticise index trackers; they have a different purpose.

Not a particularly meaningful award at all, and Eggmund is quite correct that an Index Fund would never win an award over such a short and meaningless period of time. A return of 0.09% over ten years when 5.47% is the benchmark is frankly dreadful. And then there are the fees.