You cannot fix what you never wrote down.
Most bettors know roughly how they are doing, which is another way of saying they do not know at all. A ledger with a row for every wager turns that feeling into two things you can act on: how much of what you staked you actually kept, and whether the prices you took were any good. This page covers what belongs in the log, what ROI and win rate each measure, how many wagers it takes before a number means anything, and the review habits that keep a log honest. None of it needs software you do not already have.
A bet log records decisions, not just results
A bet log is one row per wager, written when you place it rather than reconstructed afterwards. The obvious fields do most of the descriptive work: the date, the sport, the market, the side, the line, the price, the book, the stake, and the settled result. Nine columns, and between them they say exactly what you did and what happened.
Two more fields decide whether the log is worth keeping, and they are the two almost everyone omits: the line and the price at the close. Without them you can only grade outcomes, and outcomes are the noisiest thing in the whole file. With them you can grade the decision. An NBA under taken at 221.5 and -110 that closes at 223.5 is telling you something quite different from the same ticket closing at 220.5, even if both games land under and both rows read as wins.
Two small habits pay for themselves later. Keep the side and the line in separate columns, because a log you cannot sort by market and by number is a diary rather than a dataset. And record the price you actually received, not the one that was on screen when you made up your mind, since the gap between those two is itself a leak worth measuring.
What the two headline numbers actually say
ROI is profit divided by the amount staked. Two hundred wagers of one unit each that finish 12 units ahead give you 12 divided by 200, so the ROI is 6 percent. Read it as six cents kept out of every dollar sent through the market. Notice what never entered that calculation: the size of your bankroll, which is a fact about how much money you have rather than about how well you bet.
Now set it against win rate. Flat one-unit staking on NFL sides at -110 with a 55 percent record produces 55 winners worth 0.909 units of profit each, which comes to 50 units, against 45 losers costing a full unit apiece. That is 5 units of profit per 100 wagers, an ROI of 5 percent. So the 6 percent ledger above is slightly better per dollar than a 55 percent record at standard pricing, because ROI can see the prices and a win rate cannot.
That is also why 5 percent is a bigger achievement than it sounds. Break-even at -110 is 52.38 percent, so a 55 percent record is 2.62 percentage points of edge, and the return is that edge multiplied by the decimal price of 1.909, which is where the 5 percent comes from. Points of edge above break-even are the entire source of the return, and the price decides what each point is worth. The same logic runs the other way on longer prices, where a losing record can be a winning ledger, which is the subject of why winning picks can be bad bets.
Where a ledger quietly lies to you
The first lie is the denominator. Take that same 12 units of profit and divide it by a 100 unit bankroll and the log reports 12 percent, or by a 50 unit bankroll and it reports 24 percent. The wagers never changed. Bankroll growth is a real and useful figure, but it answers a question about how aggressively you stake, so quoting it as ROI makes your performance look like whatever your unit size happens to be. ROI is profit over turnover, always.
The second lie is a win rate quoted on top of mixed stakes. Suppose a baseball bettor doubles up whenever a game feels obvious: 100 wagers at -110, 55 won at one unit and 45 lost at two units, which is the worst arrangement of that record rather than a representative one. The wins return 50 units of profit, the losers cost 90 units, so a 55 percent record arrives with a loss of 40 units on 145 units staked, an ROI of -27.6 percent. A record only describes money when every row risked the same amount and every row was priced the same, and the moment either varies, only the profit column is telling the truth.
The third lie is reading the log too early. Sixty wagers is a story, not a verdict. Separating a true 53 percent bettor from a 50 percent one at conventional confidence takes on the order of 1,000 wagers, and even 500 wagers leaves a genuine edge finishing underwater often enough to matter, which the variance arithmetic in bankrolls, units and bet sizing works out in full. The ledger version of that problem is the sharper one, because a log worth reading is segmented by market, and segmenting divides a sample you already do not have: a 300-row file cut across six markets is six 50-row samples, each inheriting the problem whole. Read a young log for the questions worth asking rather than for answers, because treating 60 rows as a verdict is how sound processes get abandoned.
Habits that make the log usable
Start on day one and record the close from the very first row. A log begun once the results looked interesting is a record of the stretch that made you start it, and the earliest wagers are exactly the ones you would need to see. Log the wagers you regret in the same breath as the ones you would repeat, because a file that only contains considered bets will tell you your process is fine while your balance disagrees.
Segment by market and by sport, and fix the review interval before you need it. A flat aggregate is very often two opposite results cancelling: a real edge on MLB totals can spend a season quietly funding a leak in NFL player props, and the combined number reports nothing at all. Then decide in advance when you look, every hundred wagers or on the first of the month, and decide what you would actually change if the segment held up. Reviewing a log after a bad night is not review, it is a search for permission.
The closing column earns its keep before any other, because the profit column is slow to say anything. A closing-price column costs a couple of minutes a day and it is the field you will most regret not having a year from now, so write the closing number and the closing price beside the ones you took, and let closing line value handle what the gap between them means. It does not replace the profit column. It just fills in sooner.
The log watches you too
An honest ledger is also the earliest warning system available to you. Stake creep, rising frequency, and shrinking gaps between wagers all appear as columns in a spreadsheet well before they register as a feeling that something has changed, and they are hardest to notice during a winning run, which is precisely when nobody rereads their file. If the log starts describing somebody you did not set out to be, treat that as the finding of the month and start at responsible gambling.
Tracking and ROI, answered
How do you track sports bets?
Write each wager down at the moment you place it, rather than reconstructing the week from a settled-bets screen on Sunday, because memory keeps the wagers you are proud of and quietly loses the rest. One row per wager, with no exception for the ones placed in a hurry, is what separates a log from a highlight reel. A spreadsheet handles all of it, so the tool you choose matters far less than whether the row gets written before the result is known.
What does ROI mean in sports betting?
ROI, short for return on investment, is the share of everything you staked that you kept as profit. A bettor who put 200 units through the market and finished 12 units ahead has an ROI of 6 percent. Because it is measured against turnover rather than against a bankroll, it stays comparable between two people who bet very different amounts of money.
How do you calculate betting ROI?
Divide total profit by the total amount staked and multiply by 100 to state it as a percentage. A winning wager at -110 contributes 0.909 units of profit and a losing one costs the full unit risked, so the profit column is not simply wins minus losses. Decide once whether pushes and voided wagers count toward the staked total, because including them pulls the figure slightly toward zero and switching between the two conventions makes your own months incomparable.
What is the difference between ROI and win rate?
Win rate counts how often you were right and knows nothing about the prices you took or the amounts you risked. ROI divides profit by turnover, so both of those are already inside the number. Two bettors can finish a season with an identical record and end up far apart, because one of them was consistently paid more for being right.
How many bets do you need before your results mean anything?
A betting record needs far more wagers than most people assume before it can be read as evidence rather than as a story. Separating a genuine 53 percent bettor from a 50 percent one at conventional confidence takes on the order of 1,000 wagers, so a log of 60 rows cannot settle anything. The same requirement applies to each slice of a log separately, so a market you have bet forty times cannot be judged simply because the file as a whole has grown large. Read a short log as a description of what happened, not as evidence about what happens next.
What should you record for every wager?
Record the date and time, the sport, the market, the side, the line, the price you actually received, the book, the stake and the settled result. Then record the line and the price that the same market closed at. The second group is the part almost everybody skips, and it is the part that lets you judge whether a wager was a good decision independently of whether it won.
Should you track closing line value as well as profit?
Tracking both is worth the small amount of extra work, because they answer different questions on different timescales. A closing-price column fills in on the day of the wager and starts forming a pattern within dozens of rows, while the profit column usually needs hundreds before it can separate skill from variance. Profit is still the thing you are trying to produce, so treat the closing-price column as the early read and the ROI column as the verdict.
Is a 5 percent ROI good in sports betting?
A 5 percent ROI held over a long sample is a strong result rather than a modest one. At -110 with flat stakes it corresponds to winning roughly 55 percent of your wagers against a break-even requirement of 52.38 percent, so it represents about 2.6 percentage points of edge. The hard part is not reaching 5 percent across 100 wagers, which luck manages on its own, but still showing it after a thousand.
We keep the whole ledger, not the highlights
Every projection we publish is stored with the price that was on the board when it went out, so performance is measured against what was actually available rather than reassembled afterwards.