Tuesday 21 November 2023

World Cup Cricket and Two-Thirds Century Wraps

Despite finishing with a loss in the final, my first real participation in cricket betting during this World Cup has been a positive one. 

As I've been mentioning these past few weeks, backing most favourites in World Cup games had a decent ROI of 4.9% from the previous three tournaments and while overall the ROI has since declined to 4.3%, by passing on the ultra-hot favourites which historically are a losing proposition, the 2023 numbers were +1.63 units from 40 matches. 
With Thanksgiving Day coming up, an American wife and a two-thirds of a century "birthday" coming up tomorrow, I'll be on Overs this week. Overseas with the in-laws, over-eating and possibly over-drinking for the next few days. I like to mark these special occasions, although my wife thinks I'm a little crazy. For my one-third of a century I went to the Bahamas, and for my 25,000 days celebration in September of 2025 - which also happens to fall on a US Holiday (Labor Day) - well who knows. Hopefully I'll still be above ground, but life can come at you fast. 

Saturday 18 November 2023


This season, the NBA has introduced an "In-Season Tournament" which isn't exactly the most exciting of names for a new competition, but is basically a competition within the current regular season whereby some games effectively count double. Wins or losses count towards the standings in the traditional 82 game regular season and also to the standings in the new tournament.

The 30 teams are split into six groups of five, three East and three West, with each team playing four games against their mini-league opponents, two home and two away. The six group winners plus one wildcard team from each region will then play a traditional single elimination tournament with quarter-finals, semi-finals and a final, with the last two rounds in Las Vegas in early December. 

As with any rule or competition format change, it's often worth paying attention to see if there's an edge to be found, but with just 73 games played so far, it's not possible to draw any firm conclusions. It's not even clear whether teams or players will be taking these doubly important games any differently. There's a nice cash bonus ($500,000) to players on the winning team which sounds like a lot of money but to the elite players on say $30million a year, it might seem insignificant. The average NBA salary is around $9.7million which is the highest for any team sport in the world, but averages are distorted by the superstars (Bill Gates walks into a diner...) and the the median salary is a paltry $4.6million this season. Four teams have already been eliminated. 

Early signs are that these games are going under more than non-in-tournament games, (the average line for in-tournament games is 227.8 versus 225.6 in regular games) and underdogs are performing above market expectations. Something to watch for.

For Sacred Manuscript followers, the Totals System is firing on all cylinders so far, and while the other main system is profitable on non-tournament games, it's negative on the in-tournament games. 

It really would be much simpler to call this new competition the NBA Cup! 

Thursday 16 November 2023

Gambler's Fallacy

I'm not going to mention any names, but one tipster with an excellent track record who I've been following for several years now, recently sent an update with an interesting, and surprising, suggestion:

We're currently a few points down for the season with both official and unofficial picks, so if you've been thinking about upping the stakes at some point, this should prove to be a great time to do that.
I couldn't agree less. Increasing stakes in an effort to recoup prior losses is loss-chasing and something which is often viewed as a defining feature in the transition from recreational to problem gambling.

If your system is profitable at level stakes, there is never any need to chase your losses. The positive expected value (+EV) will, in the long run, result in profits.

Unfortunately the long run can be very long and natural variance means that there will be sequences, often painfully long, when a +EV system will lose, but the problem with increasing stakes and chasing these losses is that we have no idea when the system will turn profitable again. 

Tossing ten heads in a row doesn't mean we should increase our stakes on tails. If anything, it would be the opposite because, if the same coin is being used each time, it may be biased, but in sports betting where the events are all independent, the system has no memory and thus no idea of when it should start reverting to the mean. 

Gambler's fallacy. Past events do not affect future probabilities.

A system may NEVER revert to the mean, because in sports betting an edge can vanish overnight, and increasing stakes would be a serious threat to ones bank. 

Moving on to the Cricket World Cup, and if you followed my 'back most favourites' strategy you'll have banked a couple more winners in the Semi-Finals with India (1.37) and Australia (1.74) prevailing. 

India are around 1.5 to win the Final this weekend and Favourites have won all three previous Finals and have a good record in Knockout games:

Tuesday 14 November 2023

Cricket and FRAN

With the opening stage of the Cricket World Cup now complete, thanks to Odds Portal I have now have the approximate prices for 167 such matches since 2011 which show that backing the Favourite in these games is a profitable strategy with an overall ROI of around 2.7%.
With the number of sportsbooks offering prices somewhat limited in earlier years, though up from five to eleven these days, the data isn't perfect but the trend is apparent and while the number of knockout games is small, just 17 so far, if we include these games, the trend becomes even stronger.

Hopefully the screenshot above is self-explanatory. It shows the outcome of backing the Favourite for both the opening group and latter elimination stages and the ROIs, so for example, Favourites in the knockout stages have an ROI of almost 21%, and in all matches of 4.3%.

All Favourites are not equal however, and the next image shows the results by band:
Backing at very short odds isn't usually the best idea, and if the shortest priced 46 favourites are excluded, the ROI climbs to 6.2%.

India are about 1.37 to beat New Zealand in the first semi-final, which is very close to the price they were to win the same match up four years ago, and lost.

Not relevant for betting purposes, but it's a curiosity that New Zealand are the only country to win a knockout game as the underdog, having done this in 2011, 2015 (both v South Africa) and in 2019 versus India and they have good pedigree in World Cups reaching the last two finals.  

When New Zealand and India met in the opening stage of this World Cup, India won by 4 wickets, but New Zealand were always in the game and at one stage were 178-2 before ending at just 273. The numbers tell me to back India though, and it's about numbers, not teams.

Another rare comment on the blog, this time from my old friend schnakenpopanz who wrote:
Hi again, As I have often mentioned, I follow your blog with great admiration. What is your opinion on rating systems? On the one hand, ELO is often used to rate teams better. I use Ken Massey's site a lot. https://masseyratings.com/ Especially for college sports, it allows for very good ratings. Thank you very much and good luck.

While I love looking at ratings, the challenge I have is using them to beat the market. A little history, but back in 1987 (36 years ago!) I bought a book called The Punter's Revenge - Computers in the world of gambling - by Tony Drapkin and Richard Forsyth (I still have it in fact) and one chapter which had a big influence on me was 8.4 "Rating the strength of football teams".  

The method was named FRAN (Football Rating Assessment Number) and was based on Elo (not ELO!) with Home teams "risking" 7% of their rating and Away teams 5%. I was quite fascinated by the idea, and spent many hours in the late 1980s / early 1990s maintaining ratings, looking for value, and with some success. 

I remember one midweek round of matches when I had a number of multiple bets on Away wins in Divisions Three and Four, (in those days you couldn't bet on single matches), and all five landed for what, at the time, was a rather nice payout. I remember it because waiting for the final results to be confirmed made me late to meet a friend at the pub, and she was less than impressed with my excuse. 

I think I've mentioned before that I'd look at the prices in the Racing Post on my commute into London, stop at a phone on my walk to the office and call in my bets which was great until my bookmaker accounts all got closed down in fairly quick succession, which was probably more than a coincidence, and made me realise that my time could probably be spent more profitably elsewhere. Similar to my reaction when told I'd have to pay Premium Charges, part of me was satisfied that the ban meant my success was being confirmed. 

But I never forgot the idea, and many years later I resumed maintaining ratings on teams which is how I stumbled across the typical profile for matches with a higher than expected probability of ending up as a Draw, which became known as the XX Draws system back in 2012 or 2013. 

Like everything else, the betting landscape has evolved a lot since the late 1980s. Betfair made it possible for me to make money consistently again from betting, but other, far more sophisticated organisations, were developing their own models to identify value, and a relatively basic model based on an Excel spreadsheet was never going to find an edge.

I think ratings compiled by others are a good starting point, but because they are widely available, you need to identify a weakness (one example might be the rating for home field advantage) and tweak the overall rating to improve it, but of course everyone else is trying to do the same thing. 

Fortunately the sports markets aren't efficient, and in some cases remain inefficient for much longer than should be expected, and if you identify new trends early, and adapt your model accordingly, there is some hay to be made, at least for a while.

"The trend has vanished, killed by its own discovery." -Benoit B. Mandlebrot

Saturday 4 November 2023

October: Texas, UMPO, Cricket, Indexes and Health

The 2023 baseball season came to an end earlier this week with yet another underdog win for the Texas Rangers who clinched their first ever World Series on the road in Arizona.

On Opening Day Texas was 50-1 to be champions, and their win is the biggest upset since 2003 when the Florida Marlins won after starting the season at 75-1. 

If you've been following along with the UMPO System, you would be quite happy with the results this season.  

36 bets, 24 wins, and a 17.96u profit while the sub-category HUMPO System had four wins from five bets, and a profit of 3.69u

ROI percentages from a single year where the number of selections is fewer than 40 are meaningless, but they are available for each season here if anyone wants to see them.

The final numbers for this season and the 19 seasons for which we have data are:
With the number of selections also low for any given week or weekend, I've decided that my system updates would be more meaningful, if not necessarily more interesting, if they were published on a monthly basis. For infrequent tournaments like World Cups, and playoffs, I'll make exceptions, but for most competitions that take several months to complete, I think once a month is more than enough. 

For October the results for some of the systems I mention were as follows:
European Football clearly isn't doing very well so far this season, and with the baseball and Rugby World Cup now over, how November plays out will be largely dependent on how the daily NHL and NBA systems perform although American Football will be contributing. 

I've also been dabbling with the Cricket World Cup where favourites appear to offer value, but not the very hot favourites. 
The above table shows favourites in six bands, from strongest to the weakest left to right, and the trend is clear. Cricket isn't a sport that I've spent much time on before, but I'm always looking for markets that might be inefficient and was inspired by this article by Paul Krisnamurty who speaks a lot of sense on political events also and is worth a follow on Twitter. I mean X.  

Overall, October was a poor month. The FTSE100 was down, as were the major US indexes, and Tesla lost almost 20% which didn't help. Of the 153 months since I started tracking, October ranked 141st in real terms, although a slightly better 131st in percentage terms. The saddest part of it is that my net worth is lower now than it was two years ago, and I'm still working, never mind that I've not yet started drawing on my retirement savings!

The three main problems people face when they retire are apparently "financial insecurity, health issues and social isolation" and October was at least a good month for the second of those categories. 

As I mentioned previously, I usually try to do Sober October, and this one was 100% successful, with the consequent weight loss of 17.8lbs, the second best ever behind the Sober October of 2021. I'm now back to the 50th percentile for my height and need to stop drinking. 

The number of days when I have alcohol is more closely correlated to weight loss than the amount of alcohol, not surprisingly a pattern strengthened by counting calories in and exercising, and on that front I set records for both steps (17,652 per day) and miles walked / run (8.63 per day). 

The winter months in blue are clearly much better for my health than the yellow summer months. Good luck in November.