Bet MMA - MMA Betting Picks

Username
Password
 

UFC Online Game

MMA Handicapper Directory
Total MMA Handicappers: 526
Profitable MMA Handicappers: 252

 
 
 
   
 

Our equations

It's important that everyone who uses the site to place bets understands what information is displayed where on the site. This page logs all changes to the equations over time and how it has affected our profitablity.

The whole point of the system is to take people's raw data (which we never alter) and be as clever as we can with it, to give the best predictions possible. There's no way I was ever going to get the equations perfect first time, so they have changed over time. I programmed the system so we can change small variables and run the whole system from the start, to see what difference it would have made to our predictions and find what gives us the best results.

The unit profit / average $10 bet profit shown around the site is the results we would have got if we'd used the current equations. Sometimes that improves a historic event's performance and sometimes it makes it worse. I try not to get too hung up on what difference it makes to historic values and try to just change the equations to what I think are fundamentally better equations. If I wanted to, I could change things so the current equation profit was about +300 units but that would be pointless if the next event's predictions were garbage.

Here are our current stats (which automatically upates).

Actual overall unit profit: + 173.80 units (ROI 20.4 %)
Unit profit under current equation: +185.36 units (ROI 28.2%)

If you want to see what we bet on at the time, you can view this page, which is linked to from here and the homepage. Also, we display the actual picks at the time, at the bottom of each event page.

Why can't you just display what you actually predicted on the event pages? We do log our actual historic predictions on this page but we can't keep the old predictions in the system. It's easiest to explain why, with an example. We give every bet a value, which is either positive or negative depending on whether the user got it right. That value then gets passed on to the user to rate the worth of their future predictions. If we make a change to the value equation, that changes every single bet value, which in turn alters the rankings for every user's predictions. That process runs from the first event to the current event, totally interlinked. So for that reason, we have to display how every user was doing under the new equations, how they affected the results and what we would have predicted.

The more events we have completed, the more data we have to work with and the better we can make the system work.

To see how our system has improved over time, you can view the changes below.

Edit #1 5th Jan 2015

Problem: People kept putting 100% result for one fighter and 100% confidence, basically just because they were being lazy or because they wanted to inflate their ranking. This was spoiling the system because we didn't know which were good picks and which were just guesses.

Solution: Edited formulae to downvote people who just give 100% (or close to 100%) picks for all their picks.

Results: Improved straight & weighted pick results. Slightly improved results on value bets. People who did this will have a worse ranking, therefore my hope is that this will force people to start doing it properly, which will result in more accurate predictions in the future.

We'd had 11 events up until this point and our predictions up to that point on that system can be found here.

Summary
Unit Profit to this point +168.55 (ROI 65.71%)

Edit #2 23rd Jan 2015

Problem: I spotted a bug whereby we were giving people credit (to their rating) for fights that ended up as draws / were cancelled.

Results: I removed this credit and unfortunately it saw a small dip in profits for the weighted predictions as we have two big win fights that were 49/51% picks that were affected by this. The changes were only to a couple of fights so no point producing a whole page of what it looked like.

Edit #3 3rd March 2015

Problem: We'd had a couple of tough events in a row and in looking into the predictions, it became apparent that some people were just guessing for some fights. I figured now was a good time to do a load of changes as we had a 2 week break till the next event.

Changes: I don't wanna detail them all as I don't want anyone copying the system but some of the changes were: Higher negative values for wrong predictions. Bonus for going against the crowd and getting it right. Some manual entry of "mitigating circumstances" for e.g. poor judging results, split decisions etc. Set a minimum amount of high quality predictions to make a unit bet.

Results: Massively increased profitability over time. Basically doubling profit in unit and avg $10 bet methods. I did get the unit profit as high as +355 but I wasn't interested so much in past performance as just doing what I think are better equations, so we've ended up around +322 units profit up from +197.53. The new equations actually made earlier event returns worse because they rely more on an accumulation of good data over time than the old equations so we had a lot of "insufficient information to place bets" on fights that we actually got right.

We'd had 17 events up until this point and our predictions up to that point on that system can be found here.

Summary
Unit Profit to this point +197.53 (what we actually predicted at the time of each event, so using the results from the first equation up to event 11, then the next equation from event 12 to 17).

Also at this time I worked on adding parlays.

Edit #4 19th April 2015

Problem: Another poor set of events (4 losses and 1 profit). This was generally just bad picks by the crowd as a whole so I was just looking to change a few things to make the system as a whole better.

Changes: Increasing the importance of mitigation factors. Gone through a load of the fights and added mitigation for bad decisions, dominant performances etc.

Results: The historic profits are reduced significantly (+285 to +230). but I know this is now a better equation, so it needs doing. This is basically just down to one fight on one card moving from a big bet with large profits to no bet. It also slashed historic profits on parlays from +43 units to +32 units. I am more than confident this improves the system overall. It's not my goal to increase "what would have been" profits but to improve the system for the future.

We'd had 22 events up until this point and our predictions up to that point on that system can be found here. Note: I realised I wasn't storing our historic parlay predictions on the site. You can still view them on CapperTek.

* Edit - we have now been deleted from CapperTek, after I reported fraudulent bets. It turns out CapperTek were likely behind those bets and the whole site is a sham. That sucks cos if they ran it properly it would have been a really good site. Anyway, you can read about that nosense here (CapperTek Review), in our blog.

Summary
Unit Profit to this point +153.53 units (ROI 30.4%). Using the results from the first equation up to event 11, then the next equation from event 12 to 17, then equation 3 from events 18-22.

Edit #5 Ongoing from 24th April to Present day

Problem & Changes: This is without doubt the biggest overhaul of the equations since the project began. After the Cormier vs Johnson event (event number 26), I decided there was a serious flaw in that we'd often predict a high certainty of a certain fighter winning because everyone picked that fighter but in reality, a lot of those picks said it was a closer fight than the odds on offer. Therefore, we were suggesting small bets on loads of heavy favourites, when there was little to no value in doing so. Since the change, the system will spit out a lot less picks in general and a lot less picks on heavy favourites. As you will see if you look at our betting favorites vs underdogs in MMA page, betting on favorites in general is just not a great idea.

During these couple of weeks I have also made significant other changes.

1. I constantly asked people to not pick loads of fights per card yet so many people kept doing that. I programmed in a massive hit to their rating / value, so their predictions are massively devalued.
2. I've made it so that people's first few events of predictions are not worth as much, so we basically weight down people who get lucky straight out the bat.
3. Loads of other stuff that I'm not gunna tell you cos it's kinda the secret formula ;)

Results: Overall, the historic profits were brought down significantly but we also bet a lot less, so therefore made a significantly better ROI, which I believe is the most important value of all. I'm far from finished with these changes, so we'll see how it progresses over the next few weeks, as I improve the equations further.

 

 
 
© Bet MMA
Powered by MMA Tycoon
Site Map
Homepage
Free UFC Betting Tips
MMA Betting Blog
Betting Favorites vs Underdogs
In Depth UFC Fight Stats
List of MMA Bookmakers
Best MMA Tipsters
Next UFC Event
Bet MMA - MMA Betting Tips & Statistics