A Minimal Spreadsheet for Tracking NBA Prop Bets
Available in US
If You Can’t See It, You Can’t Improve It
I have a friend who spent three years convinced he was a winning NBA prop bettor. He was certain because he could remember his big wins and could not remember his quiet losses. When I finally talked him into building a tracking spreadsheet and back-populating six months of bets, the answer came out to a 4.8% loss rate. He was, in fact, an average recreational bettor with a strong memory for the upside.
Memory is unreliable. The spreadsheet is the antidote. A bet log that captures every wager — winning, losing, void — produces an honest record that no amount of selective recall can distort. It tells you what you are actually doing rather than what you think you are doing, which is the foundation of every other improvement you might make.
This piece walks through the twelve columns that matter, the ROI and yield maths to drop into them, the bucketing structure that turns raw bet data into actionable signal, and a monthly review cadence that takes thirty minutes and is worth ten times that in better decisions.
The Twelve Columns That Matter
Every bet log I have seen work in practice has roughly the same twelve columns. You can build it in any spreadsheet — Excel, Google Sheets, Numbers — and the structure carries across without modification. The columns earn their place because each captures information that influences review.
Column one is the date the bet was placed. Column two is the operator. Column three is the player. Column four is the team and opponent — usually written as a four-character abbreviation pair like “BOS-NYK”. Column five is the market — points, rebounds, assists, threes, etc. Column six is the line and side — “27.5 over”, “8.5 under”. Column seven is the price taken, in decimal. Column eight is the stake. Column nine is the result — win, loss, void, push.
Column ten is the price the market closed at, which is the input for closing line value. Column eleven is the calculated profit or loss for that single bet — a formula that returns positive return if won, negative stake if lost, zero if void. Column twelve is a free-text notes field for whatever context you want to capture — “rest disadvantage”, “starter ruled out”, “shopped 1.95 vs 1.83”.
Twelve columns is the right size because it is the minimum that captures everything you will want to bucket later. Fewer columns and you cannot slice the data by operator or market. More columns and you start to skip filling them in, which is worse than not having them at all. The discipline of a tracker is in the consistency of recording, not in the comprehensiveness of the schema.
Most bet logs I have seen fail not because the columns were wrong but because the bettor stopped filling them in for a stretch. Three weeks of missing entries renders the entire log statistically suspect. The realistic time investment is about ninety seconds per bet — enter price, stake, line, side at placement, return at settlement, closing price the morning after. Build the habit. Skipping is more expensive than the time it saves.
ROI, Yield and Why They Diverge
The two summary metrics every prop bettor cares about are ROI and yield, and most bettors I know use them interchangeably without realising they are different numbers calculated different ways. The distinction matters once your bet sample gets large.
ROI in the betting context is total profit divided by total amount staked. If you placed £1,000 of stakes across a hundred bets and ended £50 ahead, your ROI is 5%. The interpretation is straightforward: each pound you risked produced, on average, five pence of profit. This is the same arithmetic as the worked example where +5% EV at +110 corresponds to 5% ROI per stake — the long-run profit per dollar staked across a positive-edge sample.
Yield is the same calculation expressed slightly differently in some traditions, but in others it specifically refers to profit divided by the cumulative number of bets rather than cumulative stake. The two diverge when stake sizes vary. A bettor who bets £20 on most picks but £200 on a small number of bets will see different ROI and yield numbers depending on whether the wins clustered in the high-stake or the low-stake bets.
For most recreational bettors, ROI is the more honest summary. It captures the actual money outcome relative to the actual money risked. Yield can mislead in either direction depending on stake distribution. My own tracker shows ROI calculated as the ratio of profit-loss column to stake column, summed across whatever bucket I am inspecting.
Both metrics need a sample size big enough to be meaningful. A 5% ROI across fifty bets is essentially noise. The same ROI across a thousand bets starts to look like signal. The reason most bettor “track records” published online are not credible is that they cover too few bets to discriminate skill from variance. Every serious tracker I know carries a “bets-to-date” figure prominently and treats sub-200-bet samples with deep scepticism.
Bucketing Bets by Stat Type and Market
The unbucketed ROI number across all bets is interesting but mostly diagnostic. The actionable insight comes from slicing the same data by category. The categories I find most useful are stat type, market direction (over vs under), operator, and time-of-season.
Stat type bucketing breaks bets into points, rebounds, assists, threes, steals, blocks, and combined-stat markets. Most prop bettors discover that their edge is concentrated in one or two of these categories. The bettor who is profitable on assists and underperforming on rebounds learns more from that segmentation than from the headline ROI number.
Market direction bucketing — over versus under — surfaces a bias most bettors do not realise they have. Many recreational bettors lean over because rooting for a player to do something is more enjoyable than rooting for them not to do it, and their book of bets reflects that lean. The ROI bucket on overs versus unders tells you whether the lean is paying for itself or quietly costing money.
Operator bucketing tells you which books are giving you better effective prices. If your bets at one UK book consistently produce higher ROI than your bets at another, you are observing the line-shopping effect at the operator-pair level. The lower-ROI book may have systematically worse pricing on the markets you bet most often. That is actionable.
Time-of-season bucketing — November vs March, regular season vs playoffs, pre-All-Star vs post — surfaces drift in your edge across the calendar. NBA pricing efficiency changes through the season as the market absorbs more information. A bettor whose edge is concentrated in early-season pricing inefficiency may need to reduce stakes through the back half of the year.
A Realistic Monthly Review Cadence
The tracker only does work if you actually look at it. The cadence I recommend, and the one I use, is a monthly review that takes about thirty minutes and produces three or four actionable observations.
The first ten minutes go to the headline numbers — bets placed, total staked, profit-loss, ROI, CLV. These four-or-five numbers are the equivalent of a quarterly earnings report. They tell you whether the month was a winning, losing or break-even period and how that compares to the longer trend.
The next ten minutes go to bucket analysis. Run the ROI numbers by stat type, by over/under, by operator. Note any bucket whose ROI has moved meaningfully in either direction. A previously profitable bucket that has deteriorated is worth investigating before it does more damage. A bucket that has unexpectedly become profitable is worth understanding before it disappears.
The last ten minutes go to the notes column. Read through the free-text observations. Look for repeated phrases — “shopped a better line”, “missed late scratch”, “lazy on the model”. The repetition is signal. The single most useful thing a tracker produces over time is a pattern of behavioural mistakes that you only see when you read the cumulative notes from a perspective of distance.
The monthly review does not produce instant changes in performance. It produces gradual ones. The bettor who reviews consistently for two seasons makes a hundred small adjustments that, in aggregate, look like an improvement of one or two percentage points of ROI by the end of season two. The broader maths and tools framework for NBA props sits on top of this kind of disciplined record-keeping; without the data, the maths has nothing to work on.
How many bets do I need before ROI is meaningful?
Most analysts treat samples below 200 bets as essentially noise. Below 500, results carry meaningful variance even for genuinely sharp bettors. At 1,000 bets, ROI numbers start to discriminate skill from luck with reasonable confidence. Below those thresholds, treat your tracker as a behavioural-discipline tool rather than a verdict on edge — the records still teach you about your own decision patterns even when the ROI number itself is not yet trustworthy.
Should I track every ‘what-if’ no-bet decision too?
No. Tracking decisions you considered but did not take adds enormous overhead and produces almost no usable signal. The actual bets you placed are the ones that affected your bankroll, and the actual closing lines on those bets are the ones that produce CLV. The bets you walked away from are not part of your performance record. The exception is when you are deliberately running a back-test of a new model — there, paper-tracking has value as a validation step before you commit real stake.
The Spreadsheet Tells the Truth You’d Rather Forget
The most uncomfortable observation about bet tracking is also the most reliable. Almost everyone who builds a tracker for the first time discovers their performance is worse than they thought. Memory edits in your favour; the spreadsheet does not. That moment is the inflection point. The bettors who keep the spreadsheet and act on what it tells them get gradually better. The bettors who shut it down because they did not like the answer keep losing money in the same patterns. The discipline is the lever; the data is what makes the lever real.
Published by the nba Props Betting team.
