One topic that I see brought up like once a month in the sports betting world is one that is actually answerable. This is a solved problem.
“Should you bet into bad/stale lines?” (knowing there is some risk of freerolling, cancellation or account closure)
I asked this question out loud in 2018 —
Turns out it’s #NotThatTricky if you really dig into the question both mathematically as well as procedurally. Whether or not you should bet into stale / bad lines is simply a function of the following —
K = (bp — (q + r)) / b
Looks familiar right! Well, it’s just Kelly, but with the added risk calculation that you need to account for that’s outside the result of the bet.
Calculating Risk (r)
Properly estimating risk requires you to factor in the following —
- How aggressive is the sportsbook when canceling bets / closing accounts — if you’re dealing with RIPTony.eu, everyone knows that Tony was super aggro in canceling wagers and closing accounts. However, if you’re dealing with a more automated skin, these sportsbooks may have a much lower “Catch Bad Line Rate”. Many of the times, the evaluation of how aggressive a book will be in canceling/closing will appear to be almost binary in nature — you’ll find yourself, in practice, determining that there are very few books that are “in the middle” on this issue.
- How popular is the bad line going to be? — If someone is spamming “omgz guyz bad line here” in a site’s chatbox that is monitored by admins, don’t even bother betting into the line.(Tangent — if you find a bad line, keep your mouth shut). If there’s say 8 Europa League games going on at once, the odds of too many people banging away at the bad line are much lower.
- The type of bad line matters greatly in the odds of detection — IMO, this is the secret sauce. This is where all the EV actually is. What type of bad / stale line will determine so much of the outcome once you start applying this approach in the real world. First, there’s the issue of betting and beating certain markets vs say props that has its own issues of getting your account flagged / cancelled that’s not really the point of this writing. Next, you must consider the type of bad line — each type of bad line will have its own detection and popularity measures. So, what are some of the different types of bad / stale lines —
- A prematch line is noticeably off market (depends — risk increases as line drifts away from widely available lines)
- A prematch line is slow / lagging onscreen moves (low risk in detection, but high risk in steam chasing detection)
- An event has started, but the prematch line is still open (high risk)
- A live line is noticeably off market (depends — risk increases as line drifts away from widely available lines, but lower risk than prematch drastically off market lines)
- A live line that’s already won but the match is still occurring (depends — how popular, how long was it available, the limits, but this extremely low on the automated detection scale compared to other bad lines)
- A live line that’s already won and the match is already over (high risk even if the line is stale for short periods, risk greatly increases as the time the event is over increases)
Many data feeds employed by sportsbooks and grading systems will include very accurate start times to matches as well as halves & periods; therefore, betting into lines still up after events have started are high risk. These processes may or may not be automated and you can generally determine which grading processes are automated by the speed of the grading. Evading detection gets far easier if you can bet into bad / stale lines that aren’t easily automated away and easier if even when a manual grader inspects the tickets for events that your doesn’t stand out.
Real World Example
Let’s give a real world example that’s the most recent stale line I’ve bet into.
This was a live, in-play soccer wager at a well known sportsbook. The event is an over/under in top tier soccer and not some prop or low tier league. There was no VAR correction that could potentially come after the 2nd goal in the 79th minute (clean header off a corner). The o1.75 live line had been up for ~2 minutes after the 2nd goal and the line stayed up for about 7 minutes. This is on a sportsbook that’s known to be aggressive canceling certain types of bad / stale lines, but they have difficulty catching stale lines in certain live events mainly soccer. This sportsbook employs autograding for soccer and they have difficulty checking timestamps for events that happen during the run of play (very good at 1st & 2nd half kickoff times though). There was no risk of freerolling since the wager is already a winner. The limits are high enough that it’s worth the risk to the small bankroll on this one out. I determined that the risk (r) for this wager was extremely low and even bet this line on multiple accounts.
The paragraph above is the thought process you need to have extremely quickly when determining to fire into a bad /stale line (or, you can attempt to automate this process …). It may seem like overkill, but avoiding the pitfalls of getting freerolled, your out closed, etc all while maximizing EV spots is how applying that process shifts the math greatly in your favor.
But What About the Ethics?
Lol.
Back in July 2020, both the Be Better Bettors podcast (episode 27 ridiculously called “Be Better People” lol) and the Bet the Process podcast (“Mystery Guest podcast) spend entirely too much time on the ethics of the above. I encourage people to listen to both podcasts to get different opinions on this topic.
But they are all overcomplicating this.
Even if you believe the ethics of any of this should even matter, which they shouldn’t, it’s still just a math equation where you’re adding additional risk to r where the additional risk is the dissolution / destruction of the relationship. However, sports betting or basically any prediction markets participation is as clearly adversarial as any activity there is in the world, but if you do have any concerns about the ethics, I leave you with this very simple advice —
