Betting as an investment

Alternative Investments
22

Well, the elections are approaching. Nate Silver has a very interesting podcast on frequently asked questions, e.g., about election forecasting, polls, uncertainty, etc. I recommend listening to it: Link to FiveThirtyEight’s podcast

Perhaps the most essential points to note are:

  • Even if the probabilities are, for example, 85-15 for Biden, and Trump were to win, this does not mean that the forecast and model were wrong.
  • Even if probabilities like 75-25 sound like “sure” wins, this is not the case: 25% is still a very possible event. Taleb (e.g., Black Swan) also speaks loudly about this topic in his books. It is challenging for people to grasp probabilities and they have an unfortunate tendency to draw such false conclusions that small probabilities = impossible events.
  • Nate Silver states that in their model, Trump’s probability of winning is currently 5%. The remaining 10% comes from temporal uncertainty (poll results and the race tend to tighten closer to the elections). So, if the elections were held today, Trump’s victory would remain very unlikely.
  • Even if the polls were as wildly off as in 2016, Trump still wouldn’t win.
  • Forecasting is difficult, and it involves a huge number of uncertainties – especially in presidential elections, of which there is a very limited historical number, and especially in these elections, which have a record number of unpredictable events whose impact is impossible to determine reliably.

I’ve continued to add to my Biden stake. Most of it is still in the popular vote, for which the probability is reportedly 99% according to The Economist and 94% according to FiveThirtyEight. In this regard, the polls would have to be completely out of whack for a win not to happen, as Biden’s current lead in the polls is around 10%. Perhaps the bigger probability and risk at this stage is Biden contracting COVID-19.

Biden’s odds have been offered at 1.72-1.5 after the last debate. It seems several strong bettors (Aki, Jens…) on Sijoitustieto’s (Investment Information) side are strongly leaning towards Biden. I also agree with Jens that the situation increasingly resembles a Mayweather-McGregor betting scenario, where the odds and realities may not be quite aligned, and there might be an overbet available.

What concerns me most right now is what impact it would have if Trump, for example, pushed through a proper stimulus package and started generously distributing checks bearing Trump’s signature to citizens. I wouldn’t be surprised if the opinion of base voters (Blacks, poor, Latinos, unemployed, blue-collar workers…) became more Trump-friendly if several thousand units of free money landed in their mailbox. People are, after all, inherently pretty stupid and greedy.

4 Likes
23

And when, in the case of presidential elections, is a “official enough” result obtained so that winnings can be paid out? Is this the day after the election (when Trump is even a likely leader), or do we wait a month for all postal votes to be counted? In a situation like this, I would want a pretty strong assurance from the bookmaker before betting, even though I am an active sports bettor.

24

Check the bookmaker’s rules or ask customer support via email. Special bets generally have different rules depending on the provider, so there’s no single comprehensive answer to your question.

25

One office states the following:

The winner will be determined by who receives the required number of electoral votes on Election Day. If a clear winner is not found on Election Day, bets will be settled as soon as one candidate can be considered to have won beyond reasonable doubt.

Despite all the chaos, I wouldn’t consider it likely that there would be any official level of messing around with the results of the US presidential election. Trump can shout his own claims, but that’s just his style. In the end, both Trump and the people will listen to the law and believe the election results, whatever they may be – we’re not in Belarus now, after all.

I see it as a positive clause that the decision isn’t forced on Election Day, but rather, if necessary, they wait for, for example, the uncounted postal votes. If these are acceptable votes according to the law, then it is, in my opinion, completely self-evident that they should be considered in both the elections and the betting. However, there’s quite a lot of whining everywhere now. This is probably just and solely due to Trump, whose strategy of maximizing chaos and confusion will hopefully soon come to an end.

26

It wouldn’t mean it, but it would strongly suggest it. If Trump could surprise twice with 15% odds, the probability of that would only be 2.25% if the model were correct. In that case, maybe one should indeed look at the model and consider if there’s some systematic bias messing things up. The polls in the previous elections were consistently wrong, giving too little weight to uneducated whites. This has now been corrected. So, nothing guarantees that there aren’t such biases this time. The lead is big enough, though, that I don’t believe this will happen.

27

The election campaign continues. I’ve been reading more about the topic, and steadily, week by week, I’ve been adding minimum amounts to Biden’s popular vote and election as the elections approach and the probabilities increase.

Currently, The Economist gives Biden a 91/9 chance of winning and a 99/1 chance for the popular vote. Fivethirtyeight also has numbers like 87/13 and 95/5.

Here are some interesting findings:

  • The Economist warned that the situation might be slightly tighter than what the polls suggest. The hypothesis for this assumption is that when your own supporter is messing up, it’s less likely that their supporters will answer survey calls.

The reason is a phenomenon researchers call “differential partisan non-response”. In less geeky words, when the candidate you support is doing badly, you are less likely to answer a pollster’s call.

  • 538 sent a survey to pollsters. In summary, efforts have been made to learn from the last election and to better account for factors that went wrong last time. This is good because pollsters can reverse engineer their models and see where they went wrong. Based on this, among other things, the survey method (completely random vs. a certain percentage of party members; phone call vs. online) has been modified, and various parameters (education, race) have been taken into account more comprehensively to better reflect the demographics of the polling area. Experts expressed concern that this might be fighting a past war, i.e., changing things that shouldn’t necessarily be touched. However, purely on a gut feeling, I believe that a) lessons have been learned from the mistakes of the last election, meaning the quality would be better this time, and b) any possible changes would favor Trump, which would further strengthen Biden’s position.

Currently, Nevada, Wisconsin, and Pennsylvania look very strong for Biden, and this would be enough for a win. Not even the margin of error in the polls should turn the situation upside down. It is worth noting, of course, that even such 10-15% winning probabilities happen once in ten, so nothing is set in stone. It would also not be unprecedented for Trump to snatch victory in these tight spots.

Florida, Arizona, and North Carolina are also slightly leaning towards Biden. Let’s hope and see if Biden manages to flip Florida in particular, which would guarantee a landslide victory. Ohio, Iowa, and Georgia are also completely 50-50 at the moment. However, I have a gut feeling that silent Trump supporters and poll margins of error would lean towards Trump, so not too many hopes in this direction.

However, there are still several weeks left until the election, and Trump is currently creating COVID clusters in places like Pennsylvania and Florida. I wonder if there are still people in the US who haven’t made up their minds?

The NYTimes reported that the situation regarding early votes looks bright for Democrats:

Huge Absentee Vote in Key States Favors Democrats So Far: TargetSmart, a Democratic data firm (unbiased?), said his models showed Democrats with a 10-point advantage among the 275,000 first-time voters nationwide who had already cast ballots and an 18-point lead among 1.1 million “sporadic voters” who had already voted. At the same point in the 2016 cycle — model showed Democrats with a 1.6-point advantage among sporadic voters.

“Democrats are highly engaged, and they’re turning out,” Mr. Bonier said. “Republicans can’t say the same.”

No one is surprised that Democrats vote more actively early and by mail.

Currently, the situation to me looks like the public will vote very actively, which in my opinion favors Democrats. The Republican approach seems to be increasingly strong, loud noise and a “make the liberals cry” mentality. I myself do not believe such an election campaign will gain support from large masses. However, because of the 2016 elections, I find myself being very cautious in both bets and statements. The situation for election forecasters and pollsters is hardly any easier.

4 Likes
28

Matched betting is a separate issue, and we won’t go into that. If one aims to make a positive return from betting, one should focus purely on value bets, not on who is most likely to win. This seems to be a common misconception for many people who consider the matter. Another common problem is looking at the outcome after the fact and drawing conclusions (solely) from that.

Let’s break it down simply. Take an example of a coin toss where person A pays B €10 if heads comes up, but B pays A €20 if tails comes up. There’s no need to explain why this is a good bet for A and a bad one for B. Let’s assume B, however, agrees to the bet, the coin is tossed, and heads comes up. So, B won. Did the bet suddenly become good for B? No, it didn’t. The bet is still bad for B, but this time, chance just went that way.

Let’s change the example slightly. The coin is tossed ten times, and if 7 heads or more come up, A pays B €10. If 0-6 heads come up, B pays A €20. The bet is still favorable for A because even though he loses the bet with a 60% probability, his expected value is positive, while B’s is negative. The bet would be neutral if, under these rules, A put €60 into the pot and B €40. If the ratio is higher, the bet is good for B; if it’s lower, it’s good for A.

This is why, for example, in the Trump vs. Biden bet, one shouldn’t think about who is more likely to win (which is quite clear anyway, it’s Biden). Instead, one should consider how much more likely Biden’s victory is in relation to the offered odds. One shouldn’t make such an assumption blindly, but for the sake of the example, let’s assume that 538’s predicted probability is perfect, meaning Biden’s winning probability is 87%, for which the odds would be 1/0.87, or about 1.15. At these odds, the expected value is neutral, and anything lower is negative. So, for the bet to be sensible, one should get at least 1.16 (+ the bookmaker’s margin), meaning it’s not worth betting at odds below 1.20 if you trust 538’s forecast. Currently, Biden seems to be getting something like 1.4-1.5, so this is a very clear value bet, IF you trust the polls.

I myself have bet on Biden for precisely this reason. This is not based on my belief that Biden will win, but because I consider the bet to be mispriced.

17 Likes
29

Should I place a small bet on Trump winning? Then I’d be a sure winner no matter what. Either I’d win a small comforting sum of money if Trump wins, or in the opposite case, I’d win valuable peace of mind when Trump stops stirring up world politics.

1 Like
30

I was expecting a critical approach, but this was completely in line with my own view. 4 days ago (as well as in my previous messages) I mentioned the overvalued odds and the comparison to the Mayweather-McGregor fight: Trump has a lot of hype and visibility, but based on statistics and facts, the odds are not favorable for him.

Both my gut feeling and the 538 forecast go hand in hand in that I give (perhaps too) significant weight to the uncertainty of the coming weeks. I still have a slightly, humanly improbable black swan fear in my ass that brings caution: What if the elderly Biden does get corona before the elections or ends up blabbing something (I feel like a victim of election propaganda, as Biden already excellently showed in the election debate that he is in full control of his faculties), not to mention if Trump pulls a rabbit out of a hat and promises the moon?

It has now been seen that 538’s forecast moves daily by about a percentage point in Biden’s direction, and thus my investment in him has also grown: As long as the odds are around 1.5 and the probabilities of these uncertainties and black swan factors disappear, Biden is, in my opinion, a profitable target with a positive expected value. In retrospect, I think it would have been more profitable to invest more at odds of 1.8 in the summer, when all statistics pointed in the current direction regarding corona, elections, and unemployment, and mainly these emotional factors caused fear.

I have, however, placed a large part of my bet on the popular vote, as the earlier odds of 1.25 (and why not the current 1.12 odds) have a quite positive expected value when one considers that Clinton already won this bet in the last elections, and polls promise a >10% lead for Biden at the national level (Trump, if anyone, can turn the right states to himself, but I don’t believe he will close that gap at the national level in a few weeks). It should be noted that this bet is for the Democratic Party, which already in the summer kept open the possibility of changing candidates if Biden fell ill.

I haven’t bet before, but this year I have put money on four targets:

  1. Mercedes’ constructor championship in F1, odds 1.18. This was a no-brainer for me, especially since the odds were still available after the first race, where Mercedes proved to be seconds faster than other teams. The only challenger here was RedBull, but the reliable Hamilton and Bottas are a significantly better duo than the boldly driving Verstappen and the inconsistently weak Albon. The season is halfway through, and it would require Mercedes to be disqualified for this bet not to come true.
  2. Democrats’ popular vote (1.12-1.18)
  3. Biden’s victory (1.91-1.48)
  4. Democrats’ House victory (1.12) <= due to a small diversification, as the probability of Republicans taking both the Senate, the presidency, and the House is probably close to zero and would be genuinely worrying. I confess that I have not calculated the extent to which the results of the presidential elections, the Senate, and the House correlate.

I probably won’t be placing any more bets in the coming years, unless there’s a clear overvalued odd in a field I understand or otherwise extremely convincing justifications. It’s hard to imagine that even then I would start betting on, say, tennis or golf, as it would then be pure guessing, which I don’t do.

It’s interesting to note that the further the current elections progress, the more, more certain, and better data is available to support Biden’s victory, but at the same time, emotionally, the uncertainty about my own bet grows (thanks to all the crap and shouting online), and it’s starting to turn from an original “investment” with a positive expected return into unpleasant gambling. If only I could bother to code my own model that would make purely rational decisions for me.

3 Likes
31

I’ve been considering this, though more to secure my current Biden position. At odds of less than 3, I wouldn’t need to put in much effort to secure my back in case of a surprise. I wouldn’t have to worry much about the election results if Biden’s average odds are around 1.75 and Trump’s 2.8. Of course, there’s still the risk that Biden’s health fails and Harris comes and takes the victory…

32

Betting would indeed be an interesting concept, but profitable operations would surely require a lot of time and dedication, which is not possible for me at the moment, so I’ve stayed away from it. Occasionally, however, opportunities arise that truly feel like no-brainers. The clearest case is McGregor - Mayweather.

I believe there should also be an explanation for these overvalued odds. If you can’t find one, I wouldn’t trust the analysis that it’s an overvalued odd. As far as I understand, professional bettors typically place bets only shortly before the event, because the whole thing is based on quick capital turnover, just like active stock trading. So, it’s not worth tying up capital for a long time even in a winning bet. Thus, especially a month or so before the event, there can be air in the odds.

A typical overvalued odds situation, as I understand it, is an event involving an exceptionally well-known (popular or hated) person who, above all, gains popularity even outside the typical follower base. Many people knew McGregor who don’t otherwise follow combat sports. In addition, he had an exceptionally compelling story, so there were plenty of recreational bettors. When the fight was announced and garnered so much hype for so long before the actual fight, the odds became truly strange before professional bettors jumped in, though even they couldn’t ultimately buy out the overvalued odds.

With Trump, it’s somewhat similar. Trump divides opinions and has many fans even in circles that are not typically very interested in politics, meaning there’s a lot of money from recreational bettors involved. In political betting, as I understand it, things are also different because elections happen so rarely that political targets can’t really be bet on professionally. Professionals focus on sports, where capital can be turned over quickly, so there isn’t the same pressure for odds to correct themselves as in sports.

Edit: If you wanted to hedge with Trump, isn’t the odd pretty bad for that? Can’t you simply hedge on Betfair by profitably selling some of the money previously bet because the odds have changed? I haven’t really delved into how this works, but I’ve understood that it’s possible.

Edit.2: Oh, and regarding Biden - Trump, doesn’t the 538 forecast go something like this: currently Biden 95% - Trump 5%, but because there’s still time until the election and there’s a non-zero probability of something unexpected happening, e.g., Biden dying of corona, the uncertainty caused by that is added to the probabilities, leading to an election day forecast of 87-13. Every moment that Biden doesn’t get corona or nothing truly exceptional happens, the uncertainty decreases as election day approaches, and the probabilities creep towards 95-5.

1 Like
33

Haven’t there been some worrying news about Biden’s presidency lately, if they are true?

34

That is certainly true.

This is also true. I recall Aki or Jens from Sijoitustieto (Investment Information) noting that they feel 538 is a bit cautious in its forecast. Due to the last election, forecasters naturally want to be more cautious than before (an assumption supported by Nate Silver’s statements and 538’s polling interviews), and in my opinion, they place too much value on uncertainty. Of course, Trump can surprise, and there’s COVID, unemployment, BLM, and more, but my gut feeling says that this significant uncertainty is more human and emotional fear and caution than genuine uncertainty and the probability of a black swan event. What is the actual probability that Biden will fall seriously ill with COVID in the coming weeks, dementia will strike in the next debate, or Trump will pull a rabbit out of a hat, especially when the news, campaign performance, poll data, etc., of the past weeks and months have been quite favorable for Biden?

This situation has the same emotional flavor in many ways as the Mayweather fight. Data and logic say that the winner is very likely Biden (/Mayweather), but subconsciously, the hype wave and loud fan base make one question the belief in their own analysis. In my opinion, this is why inflated odds are available. With Formula 1, the situation was easier because there was no shouting or trolls trying to get under one’s skin.

35

What news? Could you clarify?

36

Hi. Great discussion here. My first post on the forum.

I’d say that in addition to probabilities, psychological factors greatly influence election odds. According to my own research, even in sports betting, small odds generally don’t attract people to bet as much in proportion to their probability of occurring. This leads to betting companies being able to slightly increase the odds for the favorite and decrease them for the underdog accordingly. It’s psychologically more challenging to try to win 30 euros with a 100 euro bet than to win 100 euros with a 30 euro bet. The win doesn’t bring much satisfaction, but the loss stings a lot. As a result, systematically betting on big favorites in every match at a betting company that always offers the highest odds could even lead to a slight profit. In elections, other psychological factors are even more pronounced as views on candidates are so sharply polarized, and many want to bet on their favorite’s victory, so it’s not surprising that the odds don’t settle very close to the probabilities.

Furthermore, many (like me, with 1.20 odds and an 88% estimate) who bet on Clinton’s victory in the 2016 elections are wondering if they dare to bet against Trump again if the polls are as far off as last time. Conversely, many who bet on Trump’s victory then will gladly bet on it again, perhaps without much regard for the odds. At least the polling companies have strongly assured that there shouldn’t be the same kind of error in the numbers, and voters’ socioeconomic statuses, etc., have now been better accounted for.

But in my opinion, those 1.50 odds for Biden are crazy good odds. I haven’t touched them yet because I own quite a lot of renewable energy + hydrogen company stocks, so losses if Trump wins will already come through them, and I already have enough risk in that sense.

4 Likes
37

Hi. This is partly true. For a bookmaker, the outcome of matches is, on the other hand, irrelevant. The most important thing is that the betting turnover is evenly distributed across all selections in terms of refundable winnings.

However, systematically emphasizing favorites can lead to the betting turnover becoming polarized, meaning it leans too heavily towards favorites. Often, people also like to bet on low-odds favorites as accumulators with several selections. For example, Real Madrid x Man City x Bayern Munich, where if all favorites win, a larger amount of money is returned to the players. With big favorites, even a 0.05 change clearly affects the final odds if they are played on a multiple-selection slip: 1.20 x 1.20 x 1.20 = 1.73 or 1.25 x 1.25 x 1.25 = 1.95.

In addition, odds can also be weighted based on locality. For example, in a Finland - Norway national team match, Veikkaus’s odds for Finland to win could be lower than at Norway’s local Norsk Tipping. The assumption is that local fans will bet on their home country’s team anyway. Fortunately, there is still competition in the betting market, so monopolies cannot offer just any terrible odds to kiosk players.

This is how it works for the casual bettor, but for an amateur/professional bettor, the odds alone do not matter in terms of whether they bet on a selection. The most important thing is simply to bet on an “over-odds” selection (ylicerroin). The tables below provide a simple example of how odds affect profit and that even small odds are worth betting on in the long run (as long as they are “over-odds”).

The assumption is that the odds for ten selections played were 1.20 and the estimated probability of winning for each selection was 90%, which also materialized. Thus, 9 out of the 10 selections hit, generating 1080€ with odds of 1.20. A total of 1000€ was used for the ten bets, so an 8% return was achieved on the invested money.

|Bet No.|Odds|Stake|Win|\n|—|—|—|—|\n|1|1.20|100 €|120 €|\n|2|1.20|100 €|120 €|\n|3|1.20|100 €|120 €|\n|4|1.20|100 €|120 €|\n|5|1.20|100 €|120 €|\n|6|1.20|100 €|120 €|\n|7|1.20|100 €|120 €|\n|8|1.20|100 €|120 €|\n|9|1.20|100 €|120 €|\n|10|1.20|100 €|0 €|\n|||1 000 €|1 080 €|\n\nBy betting on an “under-odds” selection of 1.10, there would be a slight loss with a 99% return.

|Bet No.|Odds|Stake|Win|\n|—|—|—|—|\n|1|1.10|100 €|110 €|\n|2|1.10|100 €|110 €|\n|3|1.10|100 €|110 €|\n|4|1.10|100 €|110 €|\n|5|1.10|100 €|110 €|\n|6|1.10|100 €|110 €|\n|7|1.10|100 €|110 €|\n|8|1.10|100 €|110 €|\n|9|1.10|100 €|110 €|\n|10|1.10|100 €|0 €|\n|||1 000 €|990 €|\n\nThe fair odds are calculated: 1/(winning probability) = 1/(90%) = 1/(0.9) = 1.11111.

1 Like
38

Wasn’t the return on investment (1080-1000)/1000 = 0.08 = 8%?

1 Like
39

Yes. There was a typo.

40

Interesting discussion about voting issues at 538.

I wouldn’t have believed the amount of crap associated with these elections:

  • In Texas, 1 polling station per county is sufficient. The largest county in that state is the size of Delaware, where as a result, less than 30% of residents can reach a polling station in under 20 minutes! One might exclaim, “Are they trying to suppress voting here?!” The same law is also in force in Georgia, among other places. Anyway, these states are unlikely to determine the winner.
  • In South Carolina, the Supreme Court is dithering back and forth on whether a witness is required for a ballot. Some people have already voted early with instructions stating in black and white that no signature is required. Now the Supreme Court has ruled that it is required. The result is the rejection of ballots. The same issue is in court in Arizona and Georgia, among others. Of these, SC and AZ are swing states.
  • Similar signature and other problems have occurred with early voting everywhere. Problems have been common, especially among black people and first-time voters, and as is known, Democrats have been the most enthusiastic early voters so far, so the negative impact may be greater for them. I try to comfort myself with the thought that, as I understand it, only a little over 10M votes have been cast. Proportionally, this is still several percent of the total.
  • In Florida and Virginia, systems crashed.
  • Many lawsuits regarding the voting deadline. It has not yet been decided whether a mail-in ballot will be accepted based on the receipt or the postmark, etc., so it’s impossible to know how many votes will be rejected as late.

In all of this, it’s important to note that 60-70% of votes are expected to come by mail in these elections (due to corona)! So it’s no wonder that, regardless of the outcome, there will be a significant aftermath regarding the result (unless Biden manages to win by a clear margin).

Here I again bet a day’s salary on Biden with 1.5 odds. Trump’s courtship of suburban women looked so desperate that it left the impression that the campaign team might be very desperate.

2 Likes
41

This seems like a real “tracksuit system” (Finnish idiom for unsophisticated/amateurish), but it might even be a profitable way to play these days. At least it’s much less losing than systematically playing underdogs. After the Bosman ruling in 1995, which liberalized player transfers, the internal differences in skill levels in Europe’s major football leagues have grown; big rich teams have succeeded and simultaneously become even richer, emptying even the weaker teams in the top leagues of their star players.

But this same undervaluation of small odds also applies, based on my own research, to at least tennis.

It seems to affect at least one professional bettor. :wink: My betting records include 27,000 single bets, of which about 700 have odds below 1.50, meaning less than 3 percent. Somehow, selections with more even odds are more enjoyable to play. Of course, I can only speak for myself.

7 Likes