Do You Lie to Yourself About Your Losses Before You Lie to Anyone Else?

[SharpEddie47]

The spreadsheet is supposed to prevent this.

Every bet documented. Date, selection, stake, odds, result, CLV, reasoning notes. The record is supposed to be the honest version that my memory isn't.

I went back through five seasons of reasoning notes last month looking for something specific. What I found instead: the notes written immediately after losses are systematically different from the notes written immediately after wins in a way I hadn't noticed before.

Win notes: short. Selection was correct. Model confirmed. Result as expected.

Loss notes: long. The reasoning section doubles or triples in length. Explanations proliferate. The referee decision that changed the game script. The injury that wasn't disclosed until after the match. The weather that affected the passing game in ways the model doesn't capture.

The loss note that runs to three paragraphs of explanation is the internal lie in written form.

Not fabrication. Every individual element is true. The referee decision happened. The injury was real. The weather was adverse.

But the three-paragraph explanation is generated by the loss. The same match, won, would have produced four sentences.

The explanations aren't false. They're selective. The conditions that explain the loss are found after the loss has occurred. The same conditions before the match would not have been identified as bet-disqualifying.

The loss note as retroactive edge preservation. The bet wasn't wrong. External circumstances produced an incorrect result.

Twenty years of this. In writing. Documented.

The spreadsheet was supposed to be the honest record.

The spreadsheet contains the self-deception in more legible form than the memory does.

 

[FadeThePublic]

The categorization system is where I've found the self-deception most embedded.

Every loss goes into one of three categories in my records.

Category one: public got lucky. The fade was correct, the public money happened to win through variance.

Category two: line movement warning missed. The steam or the reverse line movement indicated something I should have picked up before placing.

Category three: genuine analytical error. The read of public distortion was wrong.

The proportion of my losses in each category across ten years of records.

Category one: 47%.

Category two: 31%.

Category three: 22%.

The honest question I've been sitting with: is the categorization accurate or have I been filing losses into categories that preserve the methodology's reputation at the expense of honest accounting.

The category one loss: completely exonerates the analytical process. Public got lucky. Nothing to learn. Nothing to change.

If 47% of my losses are category one: the methodology was right nearly half the times it lost.

The alternative reading: I've been categorizing borderline losses into category one to avoid examining whether the fade was justified in the first place.

The categorization system designed to learn from losses: possibly being used to explain them away instead.

 

[TaffyTipster]

The near miss accounting.

Backed Wales to win.

Wales drew.

The internal narrative: Wales were the better team. Should have won. One moment of quality from the opposition and we drew instead of winning. Practically a win.

The actual accounting: lost the bet.

The internal narrative that turns a loss into a near-win happens before I've left the room where I was watching the match.

Not a deliberate reframe.

Just the natural way the result gets processed.

The gap between "backed Wales to win and Wales drew" and "Wales should have won and were unlucky to draw" is the self-deception gap.

Both sentences describe the same result.

The second sentence has been through a processing step that the first hasn't.

The processing step happens automatically.

I've never caught it happening.

Only ever noticed it was there by reading back something I'd said after a match and realising the result had been translated somewhere between the whistle and the words.

 

[DublinDegen]

the internal lie came first always...

the social lie was just the external version of something already well-established internally...

the specific mechanics of how it worked...

a bet lost... the immediate internal response: not "i lost that" but "that shouldn't have lost"...

the shouldn't have is the first lie...

the bet existed in a world where it should have won... the result was the deviation from the correct world... not the bet...

this reframe happened in seconds... before the app had finished updating the balance... before there was anyone to tell a different story to...

the internal version: almost always constructed before the external version was needed...

so when someone asked how the betting was going the answer wasn't a lie constructed for them...

it was a report of the internal accounting that was already in place...

which made the social lie feel honest... because from inside the internal accounting it was honest...

the self that was being reported to others was the self that had already done the accounting in its own favor...

 

[ParlayPrincess_88]

The parlay legs I remember versus the parlay legs that were actually there.

A five-leg parlay dies on leg four.

What I remember: four correct legs. Got unlucky on the last one.

What the record shows: the fourth leg was the weakest selection in the parlay. The one I added because the other four felt strong and five legs at these odds would be worth it.

The leg I remember as "bad luck" is the leg that was always the riskiest part of the construction.

The self-serving memory: emphasizes the four correct legs. Minimizes the quality of the failed leg.

The accurate memory: I added a weak leg to a solid parlay and paid for it, and the weak leg was the one that failed, and this is the expected outcome not a bad beat.

How many times I've told the four-correct-legs story versus the I-added-a-weak-fifth-leg story: not close.

 

[CoachTony_Bets]

Film review is the coaching practice most directly relevant to this question.

Bad coaches watch film and find reasons the loss wasn't their fault. The officials. The bounce of the ball. The opponent's exceptional performance on one specific drive.

Good coaches watch the same film and find the decisions that contributed to the outcome regardless of what else happened.

The same events on the same footage. Different processing.

I've been in both categories across my career.

The game I lost and watched the film three times looking for the correct play that produced an incorrect result: I found it every time. Three different plays. Three different explanations for the same loss.

What the film actually showed: a sequence of small errors that compounded. No single play to point at. A gradual accumulation of imprecision that the opponent converted.

The three explanations I found: all accurate as individual observations. None of them the honest account of why we lost.

The loss that has no clean explanation requires accepting that the process was imprecise rather than that specific external events caused it.

The loss with the clean external explanation: much easier to close the file on.

Coaches who can only close files on the second type: they don't get better.

The betting equivalent: the same problem at the same cost.

 

[Klaus]

The model provides a specific form of self-deception that manual record-keeping doesn't.

When a model-generated selection loses: the loss can be attributed to the model rather than to judgment.

The model predicted X probability. The X% scenario didn't occur. The model wasn't wrong; the low-probability scenario happened.

This attribution is sometimes correct. Models produce losing bets in the normal course of operation and the loss says nothing about the model's accuracy.

The self-deception: the same attribution applied to losses that should prompt model examination rather than variance acceptance.

The model that loses on a specific type of match three seasons in a row: the losses might be variance, or they might reveal a systematic model error in that specific match type.

The examination that would distinguish variance from error: genuinely difficult and genuinely uncomfortable.

The easier path: attribute to variance and continue.

The selection frequency increase I mentioned in the previous thread: 23% over fourteen years.

Whether I've examined whether some of that increase represents genuine edge identification or unconscious model modification to generate more selections: I've examined it once.

The examination took approximately forty minutes.

The result: inconclusive but probably fine.

Forty minutes of examination for fourteen years of potential drift.

The examination designed to find a problem that I'm not sure I want to find.

 

[oli_sussex]

The note written after a loss is written by someone who knows the outcome.

The note written before a bet is written by someone who doesn't.

They are different documents.

Comparing them reveals the self-deception without requiring any other analysis.

 

[SharpEddie47]

Oli's comparison between the pre-bet note and the post-loss note is the diagnostic I should have been applying and haven't.

The reasoning captured before the event: what I actually believed.

The explanation generated after the loss: what I needed to believe to maintain the framework.

The gap between them: the internal lie in measurable form.

I have both sets of notes for every bet across five seasons. I've never compared them systematically in the direction Oli is describing.

I've compared post-win notes to pre-bet notes: reassuring convergence. The analysis was correct, the result confirmed it.

I've compared post-loss notes to pre-bet notes looking for model improvement signals: useful exercise that I run occasionally.

I have not specifically compared the explanatory content of post-loss notes to the predictive content of pre-bet notes to measure how much explanation was generated by the loss rather than anticipated before it.

That comparison would produce a specific number. The proportion of post-loss content that has no pre-bet equivalent.

That number would be an honest account of the scale of the internal lie.

I've been avoiding calculating it.

 

[ThePuntingProf]

The internal accounting that precedes external presentation is the one I've spent the most time examining honestly and I want to describe a specific mechanism that I don't think has been named precisely in this thread, which is what I call the counterfactual bet, which is the bet I would have placed if circumstances had been slightly different and which I use to evaluate the quality of the bet I actually placed, the counterfactual bet operates like this: I placed a bet, it lost, and I then construct a version of the decision-making environment in which I would have known the information I have now and ask whether I would have placed the same bet under those conditions, and the self-serving version of this exercise always finds that yes, I would have made the same decision even with slightly better information, which is the internal lie taking the form of a thought experiment, because the thought experiment is being run by the same person whose decision is under examination and the same person who has a strong interest in the conclusion being that the decision was correct, the honest version of the counterfactual bet would be conducted by someone who didn't make the original decision and doesn't have a stake in the outcome, which is why I found it useful when Margaret was alive to describe losing bets to her and ask whether she would have placed them, she would ask a small number of specific questions and her conclusion was not always the same as mine, and the cases where her conclusion differed from mine, where she thought she would not have placed the bet and I thought I would have placed it correctly even knowing what I know now, those are the cases I've come to identify as the most likely candidates for genuine self-deception rather than honest variance acceptance, and without her to run the external version of the counterfactual I've found the internal version drifting back toward the version that finds the decision was correct regardless of outcome, which is the direction the internal version always drifts when there's no external check to pull it back.

 

[TaffyTipster]

The external version of the counterfactual.

Prof describing Margaret as the honest counterfactual check.

Someone who didn't make the decision and doesn't need the decision to have been correct.

The equivalent I have is Bronwyn.

Who I don't tell about most bets.

Who therefore can't run the external counterfactual.

The honest check I have available: not being used because I'd have to explain what the bet was in order to use it.

The external check requires transparency to function.

The transparency I've withheld from Bronwyn has also withheld the check that would make the internal accounting more honest.

The self-deception and the secrecy: connected more directly than I've acknowledged.

 

[ParlayPrincess_88]

The staking memory problem.

I remember wins and losses in terms of the result.

I don't always remember wins and losses in terms of the stake.

A ten dollar win and a forty dollar loss in the same weekend: net down thirty.

What I remember: won one, lost one.

The framing as one-and-one rather than as net-down-thirty is the self-deception in the most basic numerical form.

The counting that counts outcomes rather than amounts: makes the performance look better than it is by the exact proportion that large losses and small wins coexist in the same account.

I've never tracked specifically whether my wins average higher or lower in stake than my losses.

Scared the answer would be that I stake more on the ones I'm less certain about because the potential payout is higher.

Which would mean the losses are consistently larger than the wins.

Which would mean the outcome tally looks better than the financial reality.

Which would mean I've been telling myself a story using accurate individual facts that add up to an inaccurate picture.

 

[FadeThePublic]

Princess identifying the stake asymmetry is the version of this problem that professional bettors are specifically trained to watch for.

The bettor whose wins are consistently smaller stakes than their losses: their record might show a winning percentage while their P&L is negative.

The self-deception: counting wins and losses as equivalent units when the units are different sizes.

The correct measure: P&L, not win percentage.

The reason bettors default to win percentage rather than P&L: win percentage looks better for the specific reason Princess describes.

The metric that produces the most favorable self-description: the one that gets used.

The metric that produces the most accurate self-description: the one that gets avoided.

 

[CoachTony_Bets]

The team that wins eight games and loses four doesn't necessarily have a good team.

If the eight wins are against bottom-half opponents and the four losses are against top-half opponents: the record looks fine and the team is actually in trouble.

The schedule-adjusted record is less flattering and more accurate.

The bettor's equivalent: the market-adjusted record.

Were the wins against well-priced markets or favorable odds. Were the losses in over-margined markets or at prices that already had the edge shaded against you.

The nominal win-loss record: available and flattering.

The market-adjusted record: requires work and is less flattering.

The record that gets maintained: the first one.

 

[DublinDegen]

the specific form the self-deception took for me that i haven't described precisely before...

the rolling balance...

didn't think about total losses... thought about the current balance relative to recent deposits...

if i'd deposited 200 this month and had 180 in the account: up 20 ahead on the month even though i was down 8,000 lifetime...

the timeframe of the accounting: always chosen to make the picture as favorable as possible...

losing month: this week has been better...

losing week: today's been alright...

losing session: the last bet was a winner...

the window moved to wherever the favorable balance was visible...

the unfavorable balance: always behind the window... always just outside the timeframe being examined...

the full picture: technically available at any time...

never examined except when something external forced it...

 

[Klaus]

The rolling window self-deception Conor describes is the most fundamental form because it operates on the accounting itself rather than on the interpretation of individual events.

The honest accounting requires a fixed starting point and consistent measurement.

The self-serving accounting: adjustable starting point selected after the fact.

The bet that moves the starting point to after the last big loss: the accounting is technically accurate but the reference frame has been chosen to produce a favorable result.

Every financial fraud investigation eventually finds a moving reference point.

The moving reference point in personal betting accounting: legal, common, and producing the same outcome as the fraudulent version in terms of what the person understands about their financial position.

 

[SharpEddie47]

The moving reference point and the post-loss explanatory notes and the stake asymmetry and the counterfactual thought experiment and the categorical filing of losses under variance.

Every contributor in this thread has described a different specific mechanism.

All of them producing the same outcome: a self-description that is more favorable than the honest account would be.

The methods are different. The direction they move is always the same.

The mechanisms that produce self-deception in betting aren't random errors in the accounting.

They're systematic distortions that consistently move in one direction: toward a better self-description.

Random errors would cancel out. These don't cancel out.

Systematic distortions that consistently move in the same direction: the definition of bias.

The self-deception in betting isn't careless accounting.

It's motivated accounting with a consistent direction of motivation.

 

[oli_sussex]

The honest account would be: write down the stake before placing the bet, write down the reasoning before placing the bet, write down the result after settlement, compare the post-result explanation to the pre-bet reasoning, and calculate P&L from a fixed starting point that doesn't move.

Most people do the first and last of these.

The middle three are where the deception lives.

 

[TaffyTipster]

Oli's middle three.

Write the reasoning before you know the result.

Compare the post-loss explanation to what you actually wrote.

The comparison that would be most honest and most uncomfortable simultaneously.

Never done it systematically.

My pre-bet reasoning: exists mostly in my head, occasionally spoken to Bronwyn, rarely written anywhere.

The internal lie has no written record to be checked against.

Which might be why it's never been checked.

 

[ThePuntingProf]

The absence of a pre-bet written record means the self-deception has no external check, and I want to add one final specific observation about this which is that the written record itself isn't sufficient protection because as Eddie has identified his written records contain the self-deception in the post-loss notes, and the reason the written record isn't sufficient is that writing something down doesn't make it honest, it just makes it permanent, and permanent dishonest accounting is what produces the confident self-deceiver who has documented evidence of their own favorable self-description, the honest check that the written record is supposed to provide only functions when the writing occurs before the outcome is known and when the post-outcome writing is compared to the pre-outcome writing in the specific way that Oli describes, and even then the comparison requires a willingness to find that the post-outcome explanation is different from the pre-outcome reasoning, which is the willingness that the entire motivation producing the self-deception is working against, the honest bettor who keeps records and still self-deceives is not unusual and is not failing at record-keeping, they're succeeding at record-keeping while failing at the interpretation of what the records contain, and I wonder sometimes whether thirty years of careful records has made me more honest about my betting or whether it has made my self-deception more elaborate because now when the internal lie tells me the decision was correct I have documentation to point to, and the documentation feels like evidence even when the documentation is itself a product of the same process whose honesty is in question.