25th Feb – 3rd March Bookings Code Results
Another week goes by, and the dataset is ever-increasing. Last week, I started sharing picks to a Telegram Group – we finished +3.14u for the week with 8/14 winners on main lines. Bookings predictions were shared on all games just for people to gauge which games held value. The group is still open to join, and will be until at least the March international break – feel free to join HERE.
For those in the group, thanks for joining. As you would have seen, I’ve posted stakes on main lines only for high-value games, but wouldn’t want to put you off laddering, or doubling any of the selected games. Personally I was able to bank a bit more than the 3.14u by laddering selections, with the biggest winner on the City Liverpool game where 0 cards were shown.
Value decays as you go up the ladder, but potentially consider doubling games which you prefer to compound the value.
Now, lets get into the…
Results this Weekend
- 69 picks in total
- 57 unders (82.6%)
- 40 winners (57.9%)
To summarise, it was a dreadful Saturday followed by an excellent Sunday results-wise.
With a 57% win-rate on main lines selected and stakes, almost identical to the 57.9% win-rate overall, that’s not great. The whole point of selecting games is to increase the win-rate.
There are lessons to be taken – first of all, the process of selecting games has to be improved, it has to be more meticulous and accurate (more on that later). Or, maybe volume is key.
Poor Results on Saturday
I mentioned last week how I was subtly hoping for a drop in win-rate, just to investigate any potential issues and work on improvements. So, what can be learned from this week?
First of all, there needs to be a greater weighting given to H2H meetings in derby fixtures. Derby fixtures get a bump, but it’s a flat bump. I had a look through derby games last season, and found the average number of cards increased by a certain percentage. Using a flat rate won’t suffice, so there needs to be a better method there, and that will be implemented in due course.
Or will that just overcomplicate things? I do believe that the market over-adjusts for rivalry games, with lines getting incredibly inflated. Even after just a few weeks of tracking, very few derby fixtures have gone over their line. I notably made that critique after the poor results on Saturday, but may have just been overly responsive to just some negative variance.
Second, I currently selected the bets to share based on their difference with the model. That’s a fair strategy, but perhaps too crude, so I want to investigate deeper into whether several factors can work in tandem to select a higher percentage of winners. For example, if it’s within an optimal range, in a specific league, over a specific number, THEN it’s a good bet – it needs to satisfy a number of conditions, as opposed to just one.
Additionally, as the number of data points increases, I hope that optimal range will become more accurate with time. The idea is to look into the number every single week in order to constant get a clearer image of what works best.
So I’ve noted down a few more things to keep an eye on to evaluate their precision at picking out winners. They aren’t particularly time-consuming tasks, but I do want to get it absolutely spot on, so I’ll share more on that in the write-up next week.
Price Drops
I have been posting the lists of games to the group either on the day (for midweek games with only late kick-offs), or the evening before. However, on a few occasions, prices dropped significantly. To give one example, Sheffield Utd vs Leeds opened around a 47 line, but was shared to the group at a 37 line. With a model prediction of 30.7, it was still good enough value to warrant a stake, and there were just 20 booking points shown in total but it could have been much better. Under 3.5 cards was priced at 3.25 when taken – when markets opened, U4.5 cards would have been this same price.
That begs the question – should I be posting the lists earlier? The issue with this, is referees are released at different times by respective leagues. For example, Premier League referees are released well in advance – 3-5 days. Whereas Bundesliga referees are only released the morning prior to the match.
That would mean leagues would have to be posted individually, which adds a ton of extra work. Would that be too confusing, time-consuming and risk losing accounts? Balancing the seesaw between taking advantage of value without the aforementioned drawbacks will be difficult, so will require more thought.
Again, it might be useful to get a vote on what would be better, but for now I’m going to keep posting each list by day (eg all Saturday games at once), as I have been doing. But I will make a greater effort at sending it off as soon as possible – right after Bundesliga referees are announced.
Your feedback is always welcome, and support appreciated!