Algorithms - do they work? A look back at 2019...

2019 done and dusted and time to reflect on what has been an up and down year for most.

We have seen the World Cup play out and as always a lot of couch coaches trying their luck at fantasy rugby, Superbru and the like. Something that seems to have become more prevalent this year though was the use of algorithms to try and predict outcomes.

It did bother me that none of these algorithms were being monitored for their accuracy against the bookie handicaps, in other words for the average punter they are mostly useless. Hence I came up with my own set of equations to try and quantify form, relative team strenghts and momentum and set out to let Excel do the work, tweaking here and there as I go along and with the intention to not only monitor it against other algorithms but also to give regular feedback to punters on forums such as this one.

You may have noticed my comments in the weekly rugby pages but here follows a summary of my outcomes and as measured against other freely available algorithms.

The holy grail for an algorithm is to consistantly perform in excess of 54%, in order to claim it's bookie busting. So here's how 2019 went down for the games my algorithm predicted and those same games measured against the other guys. Clearly some tourneys are more unpredictable than others!

Pro 14
Rugbyvision 26/54 = 48%
Rugby4Cast 27/54 = 50%
David Scott's StatsChat 27/54 = 50%
BettingStig 33/54 = 61%

Premiership
Rugby4Cast 20/42 = 48%
David Scott's StatsChat 21/42 = 50%
Rugbyvision 22/42 = 52%
BettingStig 24/42 = 57%

Top 14
Rugbyvision 40/84 = 48%
Rugby4Cast 42/84 = 50%
BettingStig 50/84 = 60%

Euro Championship
Rugby4cast 14/40 = 35%
Rugbyvision 15/40 = 38%
BettingStig 20/40 = 50%

Euro Challenge
Rugbyvision 23/39 = 59%
Rugby4cast 25/39 = 64%
BettingStig 25/39 = 64%

Global Totals for 2019
Rugbyvision 126/259 = 49%
Rugby4cast 128/259 = 49%
David Scott's StatsChat 48/96 = 50%
BettingStig 152/259 = 59%

Final thoughts: An algorithm is at best merely a tool, guide, or indication for a punter from which to apply his/her own nouse, gut and experience before placing any bet.

It is handy to have a reliable break even and/or bookie busting system though, especially if you need to quickly play through some bonus turnover requirements and need to therefore play as many games as possible without erroding your bonus!

I take my hat off to your courage Stig, publicly predicting these NH games.
I think there's just too much going on, not least of which is coach/manager agendas re: secondary tournaments, salaries, 6 nations, etc.

I reckon you'll hit higher numbers in Super rugby. Let's wait and see, I can't wait.

Keep up the good work tho, we are paying attention, cheers

Good stuff, any chance you could have a crack at Rugby League when it returns or would that be heresy!

Less flippantly, do you have any idea what the strike rate would have been using an average of all the forecasts, e.g. any merit in a "wisdom of the crowd" type approach?

Thanks Harry!

I have no interest in League (don't even know the rules, or are those laws also?) so I won't be spending time on that. Devoting my time to Union as it is as much a fun hobby as it is serious investment for me.

Interestingly enough combining the various algorithms and using an average doesn't yield significantly different results, they are usually mostly in the same ballpark and the same side of the line.

I think what differentiates me from the others is that they give great significance to historical head to heads between the various teams, where I completely ignore that. I couldn't give a toss what one team did to another two or three years ago or even last season for that matter. What matters to me most is what this specific team on this specific day is most likely to deliver, so I focus on current season performance, weighted towards the most recent games, and measured not ito points scored or how many points one team has beaten another by, but specifically only how each team has performed against their respective handicaps.

I also quantify relative team strenghts, cards, home advantage and so on.

I'm especially looking forward to Super Rugby!

Cheers

You putting numbers up for this week Stig ?

Happy happy New Year everyone - may it be prosperous for all of us.

There's a definite lack of interest on the forum in the NH rugby this weekend, which is not surprising given the frantic stuff of the past few months and New Year. This is also borne out by not even a topic started.

So even though this isn't the proper thread, my confidence is emboldened from last weekend, and I think I’ve spotted a few ‘Golden Gimmees’ that I’ll hang out there and shout my (humble) opinion (my confidence in brackets) from the rooftop, if anyone’s listening. I hope someone is because there's definitely money to be made here:

Premiership (everything at Sunbet):
Sarries to thrash Worcester on a revenge mission for last weekend’s loss, and easily beat the -19.5 (90%)
Exeter to continue their ruthless run on the crown by trashing London Irish and beating the -12.5 (95%)

Pro 14:
Ulster to beat the -2.5 against Munster (90%)
Dragons to slaughter Ospreys by more than 20, and beat the -5.5 (99%)
Leinster to beat the -24.5 against Connacht (80%)

There’s a few others hooting at me, but not at my certainty levels above:
Like Edinburgh to beat the -24.5 against Kings, and;
Cheetahs to beat the -5.5 against Zebre.

I've gone straight in and also placed a few accas.
Waiting for Sportingbet to put something up, but as usual they're useless. I really must get away from them.

Anything from you guys ? Anybody out there ?

Morning, am sure there will be some rugga interest this weekend, will be getting involved myself, I’ll start a thread and feel free to copy/paste your post above onto it so all can read.