Departing Thoughts

Joker Joe gets it, writing

Being off a maths bent, Kelly all the way for me. As you say, never full Kelly unless you can explicitly calculate the probabilities (as in casino games). Using a statistical model optimal kelly fraction is roughly the actual winnings over expected winnings when betting kelly stakes, and probably safest doing a fraction of that. Patience is the key.

Certainly with the XX Draws, fractional (half) Kelly would have been the more lucrative, and less stressful, option compared to the full version.

Starting on 1.Jan.2011 with a £1,000 bank, the full Kelly bets would have seen a high of £3,169.47 followed by a low of £576.96 and a current bank size of £1,079.45. Not too impressive when compared with a level stakes profit of 21.5 points. The chart for 2011-12 season (full-Kelly) is below:

By comparison, the half-Kelly would have seen a high of £1,980.21, a low of £920.76 and a current bank of £1,367.60. The chart for 2011-12 season (half-Kelly) is below:

The highlight though has been the performance of these selections in the Under 2.5 market. Here, assuming a conservative 3% edge, a £1,000 bank at the start of the season would now be sitting at £1,625.85. Losing runs have yet to exceed 4 on these, although I expect that to be broken at some point, but whereas with the draws, the variance is such that we see lengthy hot and cold streaks, (as is to be expected given the average prices), for the Unders, we see a much smoother progression. Here's this season full-Kelly chart:

Level stakes on the Unders would be 17.8 points.

Food for thought. BigAl continues to play the Riddler to my Batman writing [my comments in parentheses]:

Cassini, you write very well and your blog is an interesting read but so many of your posts are full of gross inaccuracies.

[Sentence would have been better if it had ended at 'read'.]

Firstly, the blackjack example was used to show the principle behind your statement being false. i.e. that a system can have many negative value bets but using a correct staking plan can produce a profit.

[Fair enough, but it's akin to discussing the IQ of five year olds, and suddenly saying 'but what about chimpanzees?' - Blackjack, or conditional probabilities, have no place in the sports betting debate.]

The same principle can clearly be applied to sports betting, the difference being that you won't purposely have negative ev bets.

[That is one big difference]

But if you do - and almost everyone will

[there are actually occasions when someone might do this, but for the most part no one purposely places a negative EV bet] -

then a decent staking plan can still produce a profit.

[Only if level stakes returns a profit, assuming independent probabilities of course].

If you honestly think that each individual xx draw bet of yours is value then you're living in a different world.

[If I didn't, I wouldn't place the bets. Sure, some turn out not to be value, but I don't know ahead of time what those poor bets are. You know that 'on average' each qualifier will return a certain percentage, and you go with it.]

Stick all the bets together and maybe you have found enough total value to make it profitable. Your results haven't hit the long run yet - time will tell

[Long run is subjective. Three seasons is a reasonable length of time, and with the markets changing constantly, it could be argued that much beyond three years is of limited use.]

You say "How someone's edge could possibly "always be 3% lower than he thinks" is, again, a statement that defies mathematical principles. If his value is 1%, it is 1%, not -2%! 2 plus 2 always equals 4, even in Blackjack."

Christ, you so don't get it. Read things properly man! "LOWER THAN HE THINKS" What part of that is difficult to understand.

[Mea culpa - misread that line! No need to take the Lord's name in vain though. I'm a very religious man!]

By the way, nobody knows their precise edge in sports betting - FACT.

[Agreed, on any individual game]

I might think a one-off two way sporting event is 50-50 but the price is 11-10. I think I have a 5% edge. The reality is that I will very often have over-estimated my edge. Unless you are betting on shit like roulette you don't know your edge. A professional gambler is really attempting to quantify the unquantifiable given that no two sporting events are the same.

Instead of sticking to your guns in your normal pompous way, you really should occasionally take a step back sometimes and think things through properly. Your logic and reasoning is often way off and your overall understanding of what you talk about is limited to say the least. But fair play, you have a great knack for wrapping up untruths in such a way as to convince the average reader you're correct.

[My logic and reasoning seem to work quite well for me. Pompous? Never heard that before...]

And Part 2 of BigAl's comment was this:

Three questions for you Cassini (I don't expect a reply to the above post).

[Sorry to disappoint]

When you started selling your xx draw picks how big did you think your edge was?

[Previous results / percentages as of that time, were laid out openly in the prospectus]

At this moment in time how big do you think your edge is?

[All time, the strike rate is 32.4%, 2011-12 it is 30%, 2012 is 34.5% - and the average draw price of selections is 2011-12 is 3.58, so the edge is 7.5% or so which matches up well with the ROI seen since recording the prices. Bear in mind too that with the draw price where it is, one or two wins can make a big difference on a day-to-day basis.]

Do you believe you under-estimated your edge, over-estimated your edge, or correctly stated your edge?

[The edge was correct at the time the prospectus was created. Of course it will change from one week to the next, but as shown above, the numbers are holding up fairly well]

Make it 4 actually. How have the picks done since you first advertised your service?

[To level stakes this season, the Draws are up 8.05 points, the Unders 17.8 points and the progress since the start of the season is shown in the below two charts for Draws and Unders respectively. The XX Draws were offered for subscription around September I believe - at mid-September (the 15th) the RTs were 3.95 points and -0.15 points for the Unders, which admittedly were an afterthought, albeit a logical one since the selections are looking for draws. People signed up starting around this time as I recall.

I was thinking this morning about the BTTS bet - how does the Over 2.5 goals bet in these matches work out?

BigAl's questions do remind me of a good point, which is the importance of not getting too comfortable with a system. Reviewing too often is a mistake, it's easy to overreact to a blip, but if there is an area for improvement, then it's important to act when it appears to be more than a blip. For example, after 192 selections, the French matches are performing better than the German matches, but after 15 months, the Bundesliga isn't faring too poorly, and while slightly in the red, it's nothing a couple of wins won't fix. But definitely something to keep an eye on.

And on that note, I leave you all for a couple of weeks. I shall be taking my lap-top so I may brighten your days with a post or two from paradise, although I'm not expecting betting or value or conditional probability to be high on my list. The XX Draw selections will continue without interruption. I don't see any midweek games other than the usual Friday / Monday schedule, but I'll check for any surprise re-arranged fixtures.

Aloha.