The Church of Betting

The Power of the Closing Line

The Power of the Closing Line

During the last few months, the gambling Twitter has witnessed a heated (and still ongoing) controversy. The subject of this controversy is the mythical Closing Line and its importance in betting. Since I believe there is a lot of interest in this topic, in today’s article I will dive a bit deeper into it.

We will first look at the term Closing Line, the underlying theory and why beating the closing line matters. Then we will try to answer the question, whether a successful betting strategy should always beat the closing line. The different opinions will be represented in order for the article to remain balanced. Let’s start.

Closing Line - Ups and Downs
Credit to M. B. M.

What is The Closing Line

The Closing Line is the line right before the start of an event.

For most markets, lines are available uninterrupted for days. You have the Opening Line (the line when the market opens), the Closing Line and all the lines in between.

What makes the Closing Line so special? How is it supposed to answer the question whether a betting record is the result of skill or the result of luck?We will see below. 

Financial and Betting Markets

To understand the Closing Line we must start from further away. Namely, in the domain of financial markets.

Financial and betting markets have a lot in common. Ed Thorp, a blackjack legend and co-inventor of the world’s first wearable computer (which he used to beat the roulette), started his own hedge fund after having “made it” on the gambling scene. The Kelly criterion is used for sports betting and financial investments alike. In fact, one of the richest people in the world and certainly the richest investor, Warren Buffet, is said to use the Kelly criterion for his investment decisions.

The similarities between the two worlds don’t end here. Making the line in sports betting as well as underwriting a stock listing is basically a pricing exercise. You must assign a price for an event influenced by a universe of uncertain factors. This event can be a sports game or an earnings report, but the basics remain the same. Which are the factors that affect those prices? People were asking this question ever since the financial markets existed. That is, until 1970 when a guy named Eugene Fama found the answer to this question. He found out, that…

Information Moves the Markets

In his famous paper “Efficient Capital Markets: A Review of Theory and Empirical Work”, Fama developed the hypothesis that in public markets, at any given time, all information is incorporated into the asset prices. Furthermore, those asset prices move only as a reaction to new information. Consequently, it is not possible to “beat the market” unless you are investing in riskier assets. And so the Efficient Market Hypothesis was born.

Efficient Market Hypothesis

What led Fama to the conclusion that the market is extremely efficient and incorporates all available information? Well, he argued, anyone who had information that was not known and could be used to extract positive returns would immediately start trading upon this information until he moves the price to a level where the trade is not profitable anymore. 

Weak, Semistrong and Strong Market Efficiency

There are three forms of market efficiency, each one more restrictive from the other in terms of possibilities for an individual investor to beat the market.

According to the weak form market efficiency, past prices can’t be used to predict future prices. According to the semistrong form of market efficiency, all publicly available information (not just past prices) is incorporated into the current price. Finally, according to the strong form of market efficiency, all information – public or private – is already incorporated into the current price so beating the market long term is essentially impossible.

What does this all have to do with betting?

As we said betting is very similar to finance in that you have a market for events the same way as you have a market for securities. Therefore, the market forces that tend to keep financial securities efficiently priced apply to event prices (odds) as well.

This hardly applies to soft bookmakers, since they do not represent a market in the usual sense. Soft books are rather brokers of bets. The ones who price the bets are the market maker books (a.k.a. sharps) and the exchanges. The softs for the most part copy their price from the sharps and exchanges and resell it to retail customers. 

With the sharps and exchanges, the price moves the moment a trade/bet happens. They are much closer to a theoretical efficient market. So the efficient market hypothesis applies to them quite well. The fact that they are much harder to beat only serves as an evidence of the same.

The role of the Closing Line

So far we are talking about the betting market in general. What is so special about the Closing Line? 

As we said, the markets are moved by information. The price of a security could be influenced by the potential state of the economy in, say, 10 years time. To this day, there are companies out there that have survived centuries. Not the case with sport events. They have a much shorter life span. 

If you bet on a game, the market on it will be first opened perhaps a week before it starts. Different people will bet on different outcomes and the price will start to form. The sharp book will gradually increase its limit and the liquidity on the exchange will gradually improve. Shortly before the event start you will have the highest limit at the sharp and the highest liquidity in the exchange. All this volume should have moved the line to where it is supposed to be. The game starts, the market is closed. The last line that was up before the event started is the one that incorporates all information known up to this point. Therefore, it is also the most efficient one.

From there on, the in-play line opens. The in-play line is an entirely different animal, but that is a story for another day.

You can’t win at closing prices

So there you go. The Closing Line incorporates all information about the event, which in theory makes it absolutely precise. You can’t beat it. If you only bet on closing, on average you will lose the overround.

Therefore, your only shot at winning is to bet before all information has become available. Optimally at time of line opening, maybe thereafter, but in any case before closing. If by the time the line closes the price has drifted towards and beyond where you thought it should be, you made a good bet. You managed to beat the closing line. You made an EV+ (positive expected value) bet and you will make money in the long run.

Only in this way, the theory goes, you could make money from a sharp book or an exchange. Since the closing lines are incredibly efficient, you need to bet before closing and have the closing price move below the price you have taken minus the overround. 

This is the concept of the Closing Line.

Practical Considerations

So, we have established that, according to Closing Line Theory, you win only by having EV+ relative to the Closing Line. This is also known as Closing Line Value (CLV).

Of course, that statement is based on the assumption that the theory holds. The theory itself is based on quite a few presuppositions that could be challenged. I will try to summarize those below. Some of them refer to the Closing Line Theory itself and some to using the market maker’s / exchange’s closing lines as a market proxy.

Bet Limits

For one, bookmakers have limits on their bets and exchange traders may not be able to trade due to insufficient liquidity. So, perhaps, the theory would be less true in smaller markets. In general, even the biggest betting markets are way smaller than financial markets, so they would also tend to be much less efficient compared to them.

Market Maker Taking a Position

According to the official theory, the market maker bookmaker doesn’t take a position. He tries to balance his books as good as possible by adjusting the odds in order to end up with the same exposure on all sides of an event. The reality looks different. As Pinnacle’s trading director Marco Blume has said in numerous interviews, Pinnacle profiles their players as sharps and squares and reacts to their bets accordingly (listen to an example below):

Logically, Pinnacle reacts to sharp bets much more than to square ones. Therefore, the bet volume is not the only predictor of how much the line moves, perhaps not even the main one. To the market maker it matters who made the bet. In that sense, Pinnacle’s price is not the market price, but the sharp market price, depending on whom Pinnacle regards as sharp.

Of course, square money preference is a two-way street.

Square money

A sharp bookmaker may take a position on the side for which he has a preference. However, as it often happens, the preference of the large majority of square punters maybe on the opposite side.

If too much square money is bet on the other side, eventually the bookmaker will have to take those bets into account and move the line accordingly in order to balance the book. Case in point – the McGregor vs Mayweather fight. The square money preferred McGregor, while the sharp money were on Mayweather that night. In a sense, the sharp players are the “usual suspects” who are more or less always present in a market. However, an event that attracts a lot of media attention is able to also attract volumes of square betting money that sharps can’t compensate for.

In this case, balancing the book might not be the best decision from an EV perspective. However, it might be necessary from risk perspective. In such case the closing line would not represent the true price since it would be skewed by the preference of the square players.

Line moved by an influential punter

Sometimes an influential sharp player, such as an esteemed tipster or just a large stakes player can single-handedly create a large market move by his bet. If the market is small, that move may never be compensated until the start of the game, even if the bet was not so good.

Yes, in theory this line move after a bad bet would create an EV+ opportunity. However, other players might not be ready to bet against this sharp player due to his reputation. Furthermore, perhaps a small market does not attract enough attention in order for the price discrepancy to be noticed and used by enough sharp money. 

This can happen especially if a famous tipster with a strong track record releases a pick. The decision of a single person can attract huge volumes on one side. Such decisions can sometimes be bad, even when made by a generally smart punter. In practice, that kind of “wrong way” market movement may distort the market.

Some redeeming qualities

Of course, no theory is perfect and every model is just an effort to approximate reality. How good the Closing Line Theory does that will be discussed later in the article. But for now let us look at a few practical advantages of the theory.

A very conservative evaluation criterion

Some time ago I asked on Twitter whether any tipster platform publishes the closing line for the picks of its tipsters. The user arf was kind enough to provide one such platform focused on the US market. My investigation showed that none of the tipsters there beats the closing line. 

arf noted that almost no tipster beats the closing line, therefore platforms don’t like to publish it. I agree with that statement. The closing line is very hard to beat. In other words, it is a very strict evaluation criterion, that is passed only by a selected few. If a tipster does not beat the Closing Line he might be profitable, but if he does then he surely is. And since in evaluation of tipsters the highest scrutiny is the most reasonable approach, the Closing Line can be a great tipster evaluation methodology.

You can evaluate a betting record with a much smaller sample

Joseph Buchdahl has shown that if you use the Closing Line as an evaluation criterion instead of the p-value of a tipster, the number of “good” bets you need to confirm that a betting record belongs to a skilled punter (and not a lucky one) decreases by an order of magnitude.

Closing Line vs p-value

In betting you can rarely collect tens of thousands of bets before making an evaluation. Moreover, if there was any edge in those it would have probably already disappeared until you collect them. Therefore, a p-value of a tipster’s record is perhaps not the best evaluation criteria out there. 

Not the case with CLV, where a few hundred bets would be enough. That is just another reason why the CL analysis might be the best tool to analyze tipster performance.

Empirical Evidence

Hopefully you got the point of the Closing Line theory by now. However, the really important question, which we haven’t look at till now is, does it work in practice? Let’s see.

Arbing, Technical Value Betting

In fact, there are a few proven strategies, where the placed bets systematically beat the closing line. If you are not new to this blog you are certainly familiar with them. Of course, I am talking about Arbitrage Betting and Technical Value Betting. For example, the Technical Value Betting software tool Trademate Sports collects Pinnacle Closing Lines and compares them to the advised picks. The advised picks indeed beat the Closing Line. This shouldn’t come as a surprise, but the evidence from the tool confirms the theory nicely. 

Technical Value Betting and Arbitrage Betting are all about steam chasing, hitting soft bookies where their line is too slow to update and reflect Pinnacle’s. Eventually they catch up, but hopefully you managed to land a bet in the meantime.

Doesn’t work every time

If there are no further line moves, you would have beaten the CL with your bet. Of course, till the beginning of the event many things can happen. The line can go up and down for various reasons, but on average you will stay on the right sight and would have make an EV+ bet that beats the CL. If the line move is big enough, you also get an arbitrage opportunity.

In essence, Technical Value Betting and Arbitrage Betting tools such as BetBurger or RebelBetting work according to the same principle.

Now, Arbing and Technical Value Betting are among the oldest tricks in the book. They are relatively simple and they work. This clearly shows that beating the closing line is a strong indicator of success, in practice as well as in theory.

Opposition to the Closing Line Theory

Since the theory has recently gained in popularity, naturally it also attracted some opposition. As previously mentioned, almost no tipster out there manages to beat the closing line. So, logically, a wave of criticism came from this corner of the betting community. 

The claim was that due to some practical peculiarities of the betting markets (see above), the Closing Line Theory is not always valid and in certain cases you can beat the market without beating the closing line.

Now, hearing this argument from the tipster community should not be surprising. But is there merit to it?

A practical case: Nishikori, the Tipster

Nishikori (Twitter handle @nishikoripicks) is the second highest ranked PRO tipster of pyckio and has a profitable track record of 3071 bets and 9.0% ROI as of the time of writing. As most other pyckio PRO tipsters, Nishikori on average fails to beat the Pinnacle Closing line. He is pretty close though, with -0.2% expected ROI against the CL + overround. Which, according to CL theory would put his expected value around zero, pick and operating costs not included. And yet, his long-lasting strong results can make even the strongest proponent of the Closing Line Theory question his conviction.

The main question is whether this performance is a result of luck or skill. Nishikori himself published a statistical analysis measuring the p-value for his performance. The p-value should answer the question “with what confidence can one reject the hypothesis that this performance is the result of luck”. With a p-value of 0.000035 it seems one can be pretty confident there is some skill involved here.

p-Value Nishikori Tipster

Survivorship bias

However, there is another side of this story – survivorship bias. If you analyze the p-value of a single tipster in isolation, you disregard the fact that you have picked him up from a lot of many tipsters, the majority of whom did not manage to achieve that performance. In fact, many of those probably achieved what could be an extremely unlucky run – performance significantly below their expected value based on the ratio odds taken / (closing line + margin). So, there is a possibility that you are looking at an extreme statistical outlier as opposed to an incredible display of skill.

Since the vast majority of sports bettors / tipsters don’t have an edge over the market, this possibility is much more likely than you would think. How do you account for that in your analysis?

What is your prior?

Cognitive Football suggests implementing a Bayesian prior in the calculation. You can multiply the observed performance indicator with your prior belief about how likely it is that the performance is the result of skill. Needless to say, for the average tipster this would be quite low and significantly impact the end result. Still, this is the correct way to perform a tipster evaluation. Only in this way you can account for the high probability of a false positive.

The rationale behind the methodology is explained in much greater detail Cognitive Football’s article on the matter. I really enjoyed it and would strongly recommend it (as well as the blog in general).

So, is it luck or is it skill?

As you see, the decision on evaluation methodology significantly impacts the result of your evaluation. Whether you use a Bayesian or a probabilistic methodology is partly a philosophical question. In the above case it probably doesn’t change the end result, but depending on your prior it could and in most cases it would. Whether you bet your money on a tipster who does not beat the closing line but has a long and successful track record is again a difficult decision with many aspects to be considered.

This article does not aim at answering this question anyway. It rather tries to explain the Closing Line Theory and what does the odds taken compared to the closing line tell you.

Nishikori himself admits that the CL theory does hold some value, since his bets that beat the closing line return more than the ones who don’t.

Closing Line vs ROI

The Church of Closing Line

Nishikori also rants a bit about the CL fundamentalists who do not see the other side of the story. Before renaming my blog to “The Church of Closing Line”, with this article I wanted to show you this other side too. I believe there is value not captured by the closing line. I just don’t know how to measure it, so I prefer the measurable value to it, that’s all.

Fun fact: Nishikori made a bet with BettingisCool’s Chris (Twitter handle @BettingIsCool) last year that he will achieve an ROI of at least 2%, all costs considered, in the next year. The bet was already settled in favour of Nishikori. A second bet offer followed, but this time for an ROI of at least 4%, which was refused. Meaning, both sides to this argument see the expected ROI of the tipster minus expenses somewhere between 2% and 4%, none of which is explained by the Closing Line.

It is unclear how good the estimation is, but if you believe N. Taleb’s “skin in the game” theory (and I do) it is certainly an honest one for both sides.

Оther evidence against the Closing Line Theory

The success of some esteemed tipsters who don’t beat the closing line is not the only empirical evidence against the theory. Betting on angles is another age old strategy and the fact that a few punters have been successful with it shows that there might be some fundamental mispricings in the market that are not explained by the closing odds. 

In fact, Joseph Buchdahl has found out that backing steamers is generally more profitable than backing drifters. This disproves even the weakest form of market efficiency for the betting markets. Remainder: it stated that past prices can’t predict future prices.

Another example for this would be the Premier League draws strategy, but there are certainly many more.

What do we know so far?

To summarize, there seem to be certain situations where the closing line does not capture all relevant information. In most cases, especially for bigger markets, it certainly does. You need to be aware of the theory’s shortcomings, but also must be aware of the theory itself and how to implement it to evaluate betting records. Within the next chapter I will show you how to do that.

How to calculate the EV based on the Closing Line? 

A bet has been placed and settled. Whether it won or lost is irrelevant. Whether it managed to beat the closing line on the other hand should give you way more information – according to the theory at least. You quantify this information via measuring the EV (expected value) of your bet. If the odds you’ve taken beat the closing line + overround, you have an EV+ (positive expected value bet). You did a good job, regardless of whether your bet won or lost.

How do we calculate the EV? You derive the true probability from the margin-adjusted closing line. You multiply the true probability with the odds you’ve taken. That is your expected value.

Example

Say you placed a moneyline bet @2.1. The line closed @1.9, with the opponent being priced the same. Since the closing line represents the true probabilities, the market assigns a true probability to the outcome you have bet on of 50%. Your expected value is therefore:

Stake * (2.1 – 1) * 50% = Stake * 5%

You have an edge of 5%. If your stake was 5 units, that means an EV of 0.25 units from this bet.

In general, the formula is:

Edge = Odds taken / (Closing Odds * (1 + Overround)) – 1

EV = Stake * Edge

If your Edge and your EV are positive, you have achieved CLV (Closing Line Value) and you are doing a good job.

Important note: The above example refers to a moneyline bet. In a point spread / total bet, instead of the odds moving, new information will move the line to a new value. For example, if you bet a basketball game for u170.5 @ 1.9, but it closes at u165.0 @ 1.9, you have again beaten the closing line. By how much exactly depends on what a point in basketball is worth in this league.

Actually, I came around a video uploaded yesterday that shows another example of the scenario above using a point spread. It also explains the concept of the Closing Line Value pretty nicely. I would encourage you to have a look:

Summary

The closing line theory has its shortcomings. However, it remains a strong indicator of profitability, which makes it a powerful tool for every bettor. Beating the closing line is not the only indicator of profitability, but is a very important one. I already talked about the importance of my LoL model beating the closing line. If a tipster beats the closing line in a liquid market, you can be confident in his abilities. Having a methodology that, under certain circumstances, gives you this confidence, is incredibly useful.

The closing line is also an evaluation benchmark only very few live up to. That’s why it is not being talked about much from people who have no interest in you evaluating them with it.

That being said, there are bright examples of those who present this data, regardless of how it speaks about them and their record. Beat the closing line or not, that is always a good sign. I would advise you to demand transparency from anyone you chose to buy a betting service from. See the data and decide for yourself. If this data is not available, stay away. I have seen tipsters that don’t beat the closing line, whom I would still give the benefit of the doubt, just because they are transparent.

Conclusion

I hope I managed to convince you of the value of the Closing Line Theory, its benefits and shortcomings. Also, I hope you will be able to implement this knowledge to improve your betting action. If you have any questions or remarks, I would love to read them in the comment section below. 

Make sure to stay tuned for the next article, which probably will be a review of one of the great sports betting books I have recently read. That aside, I continue working on my League of Legends model and will release a new article as long as I have made some meaningful progress. Now is the right time for that, since in a few weeks the main football leagues will begin – and with them another Arbing and Value Betting season, which will certainly swallow most of my free time. I will give my best to advance my model significantly until then, which should hopefully expand my arsenal of EV+ betting strategies. 

Until then, thank you for reading, enjoy the warm summer and see you around! 

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.