Betting Theory: Bet Early

In our Bet Often article, we explored the benefits of placing as many bets as you can, in order to reduce variance and have your performance deviate less from your expectation. Here, we'll talk about another broad change you can make to your betting which will increase your expected value!

Odds markets, like any financial market, ebb and flow depending on the behavior of market "players" - basically anyone buying or selling on financial markets, or anyone placing bets on odds markets. Thus, the odds that are offered to you on one day may not be the same the next day - this is not necessarily advantageous or disadvantageous, simply worth keeping in mind. What we've found, however, is that placing your bets as early as possible tends to give you the best payoffs over a long period of time. In this article, we'll discuss why that is, and how to think about what it means for the market to "move".


Initial market odds

To accept bets on a match, an oddsmaker has to offer some initial bet on their market. Where does this come from? When odds are first listed, there's no public market data to push the odds around, so the bookmaker has to make their own predictions. They have their own models for this, of course, but here's the thing: their predictions don't have to be perfect! Because of both the vig that they take, and their ability to shift their offered odds around to "balance their books" - that is, to make sure an equal amount of money is being bet on both sides, they can largely avoid actual exposure to the outcome of the match. This is where our edge comes in - if our model is better at predicting match outcomes than their quick initial estimates, we can get in at favorable odds and over time make a profit.

Of course, oddsmakers can't be wildly off either - if their estimate is obvious incorrect, everyone will just take the other side of the bet, and moving their odds far enough to fill the other side of the bet will almost certainly leave them exposed overall.

We'll be writing an in-depth article on market dynamics and the incentives of bookmakers in the future. For now, we simply have to believe that their initial estimates are reasonable, but not perfect - their goal is to set the line in such a way that both sides of a bet seem equally fair to the average bettor.


Market moves

A market "move" is simply any time the odds offered by a bookmaker change. Let's work with a simple example:

You want to bet on the epic clash between your favorite two contestants from these articles, team A and team B. Your bookmaker's initial estimate is that this is a 40/60 matchup (that is, team A has a 40% chance to win, and team B has a 60% chance to win). They add on a 6% vig, and offer the following odds:

  • Team A: +135
  • Team B: -176

(for a refresher on how odds, probabilities, and vig are related, read our Vig and Implied Probabilities article!)

Turns out the public largely disagrees, and most bettors place bets on team B. Now, if team B wins, the bookmaker is potentially out a lot of money (they may still be EV positive, but they'd rather just guarantee a profit out of their vig). So, they adjust the odds:

  • Team A: +213
  • Team B: -292

For any new bets placed at these new odds, the payout if team A wins is higher than before, while the payout if team B wins is lower than before. In theory, this should incentivize more bettors to bet on team A, thereby balancing the bookmaker's book.

This, in short, is a market move.


Directional moves

We don't just care about whether a market moves or not, however. We care about which direction it moves - in particular, whether it moves toward us or away from us. For this to have meaning, we have to have our own estimate of the probabilities of each team winning this match.

  • If the market moves in such a way that its implied probabilities are more similar to our estimate, we call this a move toward us.
  • If the market moves in such a way that its implied probabilities are less similar to our estimate, we call this a move away from us.

Going back to the prior example, where the market moves from +135/-176 to +213/-292: if we convert this to (normalized) probabilities, the market has moved from 40%/60% to 30%/70%.

If our own estimate of the probabilities was actually 50%/50%, then the market has moved away from us - relative to where it started, the 30% estimate for team A is now farther from what we believe.

Conversely, if our estimate was 25%/75%, then the market moved toward us. It is still more bullish on team A than we are, but the implied probabilities are closer to what we believe they should be.

What if our estimate was 35%/65%? In some sense, the new market-implied probabilities are just as far from our own as they were (5% off in either direction). However, directionally, we can say that initially the market had a higher probability on team A than we think it should, and now its estimate for team A is lower. So, even though the market move overshot us, it still moved in our direction.

Broadly speaking, we can think of the final odds on the market (at the last second before the match) to be the overall public sentiment of the outcome. If we disagree with the market's initial estimate of probabilities, then we can (somewhat loosely) say:

  • If the market moves toward us, then the public agrees more with our estimate than with the bookmaker's initial estimate.
  • If the market moves toward us and overshoots us, then the public disagrees with both estimates, but we are closer to public sentiment than the bookmaker was.
  • If the market moves away from us, then the public disagrees with both estimates, but we are farther off of public sentiment than the bookmaker was.
  • If the market doesn't move, then the public agrees more with the bookmaker's initial estimate than with our estimate.

This doesn't necessarily prove who's right or wrong - bettors often place bets on their favorite team rather than try to find mispriced odds. It's simply a useful barometer to assess whether or not our model is making reasonable predictions.


Courses of action

So you looked at the market when odds on a match were first available, made a decision, and then checked back later and saw that the market has moved. What do you do?

Here are the four possible scenarios:

  1. You placed a bet, and the market moves toward you: Do nothing! For the same bet, the market is now paying out less money if the team you bet on wins, so be happy you got in early.
  2. You placed a bet, and the market moves away from you: Increase your bet at the new odds, such that your total bet would be the new target size. Ideally you'd have gotten the whole bet size at these better odds, but there's nothing you can do about that.
  3. You did not place a bet, and the market moves toward you: Do nothing! The bet was already not worth making, and now it's paying out even less.
  4. You did not place a bet, and the market moves away from you: If the market moved far enough, it may be worth placing a bet now. In some sense, this can be treated as a brand new match, so recalculate your EV and see what it suggests.

In the two scenarios where the market moves away from you, it may seem counterintuitive to put more money down when the market is saying that not only was the bookmaker's estimate better than yours, but in fact it still didn't differ from yours by enough. If you still believe your model, however, market dynamics should not affect your confidence in your model. If the market is consistently moving away from you, that may be a good indicator that your model has some systemic issues, but second-guessing yourself on every individual bet that the public disagrees on is a good way to undermine your own quantitative edge. We talk about this briefly in our Team Liquid v Cloud9 prediction, and will expand on this in a future article.

It may also seem that constantly checking the market and adjusting bets can lead to very complicated situations, and if you flip-flop a few times you're just bleeding vig. In reality, if odds move at all, they only move in one direction. So, the two most important spots are the two endpoints: right when the odds are first placed on the market, and at the last second before a match begins. If the odds are moving toward you, you don't want to place additional bets anyway. If the odds are moving away from you, you want to let them move as far as possible before taking another action, so waiting longer benefits you. (it's not worth worrying too much about getting in at the exact beginning or end - odds tend to change pretty slowly if at all, and the most important thing is to just get as many EV-positive bets in as you can)


Bet hedging

In our first scenario (you place a bet, and the market moves toward you), while generally speaking you don't need to do anything as you've already locked in a good payout, there is actually something you can do: you can now bet on the other team to hedge your bet.

Hedging is a way of reducing your exposure to chance by betting on both sides of an event. If the market has moved far enough, this allows you to guarantee a profit in the outcome of either team winning, since you've gotten bets in at different odds on each side. However, because the bookmaker charges a vig, your guaranteed profit will always be lower than the expected value of your initial bet. So, in effect, you're trading some EV for consistency - the more you bet on the other side, the more hedged you are, up to a point where you are fully hedged. At this point, you will always get the exact same return no matter which side wins, so there is zero variance in your profit - but you'll be giving up quite a lot of potential gains for this.

We're planning a whole article on this, as there is some interesting mathematical nuance and complexity that's worth exploring. For now, however, we would advise against hedging. Markets don't often move far enough for it to be worth it, and our strategy revolves around maximizing EV, with the number of unique bets we place being our way of combatting variance.


Our model

As we mentioned above, we can treat market moves as a way to measure how well our model captures public sentiment. Again, we'll simply look at the first and last offered odds on every match, and see how often the market moves towards us.

To reduce some noise, we'll only look at significant market moves, which we'll define as at least a 5% change in the implied win probability of each team. It's worth noting that the market doesn't always go through large shifts - often the initial estimate is very close to the average public opinion, and often there's just not enough betting activity to push the needle.

Across all of our match predictions:

|           metric |    val |
|------------------|--------|
|    Total Matches |   2195 |
| >5% Market Moves |    437 |
|     Moves Toward |    324 |
|       Moves Away |    113 |
|   % Moves Toward | 74.14% |

In total, we've made predictions on 2195 matches, of which 437 had significant market moves between when the odds were first offered to the time of the match. Of those moves, almost three quarters were toward us!

Keep in mind that this is across all matches for which we made a prediction. In many cases, our prediction is very similar to the market, so these moves aren't necessarily meaningful. If we look at just the matches for which we would have suggested a bet (which necessitates us disagreeing with the market by a fair amount):

|           metric |    val |
|------------------|--------|
|       Total Bets |    959 |
| >5% Market Moves |    230 |
|     Moves Toward |    191 |
|       Moves Away |     39 |
|   % Moves Toward | 83.04% |

83% of the time that we significantly disagreed with the initial market odds and the market had enough volume to move the odds, it moved toward us!

This is a great sign - the bookmakers' initial estimates are often skewed, and are more likely to align with us afterwards. Since the market is more likely to move toward us than away from us, it is vitally important to get bets in as soon as possible! As the market moves toward us, the potential payout on the team we would want to bet on gets lower, which obviously brings our EV down.

In fact, we can quantify this difference: below is a plot, showing our performance since we first started tracking odds in mid-January through mid-May.

Betting early has a significant gap to betting late, and that gap is steadily growing over time, which means that this effect is persistent and thus likely to continue going forward.

Here's another way to look at it: if you believe your model works (as we believe about ours), you should expect the market to move toward you more often than it moves away from you. Given that, you should be placing your bets as early as possible, because your expected payout for the same bet always goes down as the market moves toward you.


Keeping up with bets

In  short, to truly maximize your potential and become an effective quantitative bettor, you should be placing your bets as early as you reasonably can. Most matches are bettable about a week beforehand, so getting in right before the match will give the market a whole week to catch up.

While this can seem daunting to keep up with, that's what we're here for! Subscribe at the bottom of this page to stay up-to-date on all of our match predictions. On our Suggested Bets page, you can find links to both our free (LEC, LCS, and select tournaments) and premium (every. single. match.) bet suggestions, and each page is updated daily. New predictions from that day will be listed on top so you can quickly see just the new suggestions, but we also keep a running list of every single prediction we've made.

The more bets you place, and the sooner you place them, the greater your potential profits, so check back daily!