BookieBashing ComboBets tracker – an updated study of p/l Sep 2021

bashertom Announcements

Mar-Sep profit/loss (betting to win £200 across all bets in each game)

Dec-Sep profit/loss (betting £100 on each bet)

The latest results of the ComboBets tracker can be found here. Betting £100 on every bet that has come through the tracker would have resulted in -£130k.

Background

I won’t lie – the last time I wrote a blog (link here) on the performance of the ComboBets tracker, the feedback wasn’t exactly sanguine, to say the least.

However recently there have been a couple more questions and I would like to address them.

So safety helmets at the ready, here we go with an updated study.

Combobets tracker

The ComboBets tracker monitors Goals, Corners and Cards bets under #YourOdds, #PickYourPunts, #RequestABets and #WhatOddsPaddys and calculates the Fair Odds of each bet.

After a successful 18 months of trialling a manual system, we integrated all the intelligence into an automatic system and released it in December 2020.

It is fair to say that the ComboBets tracker didn’t get off to a great start. The first couple of months saw losing bet after losing bet.

We don’t need this tracker at bookiebashing – we have a huge range of tools and trackers and we could scrap this one at any second without causing a significant reduction in the service we offer.

So why do we keep it?

We monitor all trackers and tools on a regular basis, and it is our opinion that it is a +EV tracker that suffered a bout of bad variance when it was first released.

As of the 1st September 2021, the maths of the poster above holds roughly true. Betting £100 on each bet on the combinations tracker would have lost a player £130,307,46. It may be easier to contextualise this in terms of ROI:

There have been 10,094 bets and they would have returned -£130k. Whilst this is technically an accurate statement, it would have been a high risk strategy to bet £100 on each bet in a game. There are an average of 6 bets per game, however many games have just 1 bet whereas some games have over 60, at odds up to 1000/1.

There have been an average of 6.00 bets per game. In the 1st 2 months of the tracker, there were an average of 9.00 bets per game. This is a reflection of changes made to the tracker and a reduction in the quantity of value from Paddy Power.

More noticeably, there has been a huge reduction in the number of games with 40+ bets in a single game. In the early days, when 1 component was good it meant that a large number of Combobets were good at that bookmaker (usually paddy power) as they would have all been priced up using the same mean.

84% of the games 19 games that had 40+ bets were in January and February. And these games account for an unequal proportion of the negative return of the combobets tracker if betting £100 on each game.

Variance

Variance of the ComboBets tracker should be measured in terms of games as oppoosed to number of bets. Most bets within a game will be inter-related. For example a game may have +EV bets for

  • 3+ goals and 6+ corners
  • 3+ goals and 9+ corners
  • 4+ goals and 6+ corners
  • 4+ goals and 9+ corners

If there are 2 goals and 12 corners in this game then all bets lose. When assessing variance it is therefore necessary to look at the quantity of games as opposed to the quantity of bets.

Betting to an equal Win liability in each game (Weighted level staking)

An alternative staking strategy for the ComboBets tracker could be to bet each game to a set liability. This brings two differences to betting £100 on each bet:

  • We normalise the liability of the odds. So instead of betting £100 on a 1.2 and £100 on a 1000/1 we bet each bet to win £200.
  • We limit our exposure on each game. If we bet £100 on every bet blindly, on some games we will be £6000 and on others we will bet £100. We are at the mercy of the games with a large number of ComboBets. Instead we can bet to win a level liability (e.g. £200) overall across all bets in a game.

To do this, we simply weight the stake of each bet by a function of the odds relative to all of the odds in each game.

We have had 1900 games pass through the combinations bet tracker so far, up until 1st September 2021. Using the Weighted Level Staking method:

  • Up to March 2021, the average daily ROI (90 day rolling) of the ComboBets tracker was -20%.
  • By September 2021, the average daily ROI (90 day rolling) had climbed to +8.6%

Alternative strategies that can maximise the returns from the tracker include using a form of Kelly Staking, and restricting the bets to those at higher EV.

Summary

At bookiebashing, we monitor, refine and analyse all of our tools and trackers on a regular basis. We are confident in the long term potential of the combobets tracker. Due to the nature of the tracker it may take a long time to get back in to profit when staking a level amount on every bet, because we had 60+ bets per game at the beginning and there has now been a significant reduction.

When betting using a weighted level staking methodology, the combobets tracker has been in profit from March 2021 to the current day.