From Pressing Audit to Pipeline: The Quiet Rebuild of Cricket Data in the Regular Season

Over the past three weeks, one small but irritating detail caught my eye. When I sorted the ball-by-ball logs of five domestic matches played at Mirpur's Sher-e-Bangla National Stadium by match ID, the death-over bowling economy had risen by roughly 0.83 runs compared to last season — yet wicket fall in those very overs was up 11 percent. On first look this seems contradictory: more wickets should mean fewer runs. But once fielding pressure and dot-ball patterns are separated out, it becomes clear the story is not a bowling failure but a measurement error. And to find a measurement error, you first return to the pipeline. A clean match ID is worth more than a clever model — I no longer say this as a flourish, I prove it in daily work. I have watched cricket matches for many years, and I have built the habit of treating every match as a data problem. In 2026, when I set up a standardized pressing and shot-location collection template for the Dhaka domestic league, I learned that the biggest enemy of messy data is inconsistent definition. I trained three interns in Khulna to log every delivery, every fielding pressure and every covered distance separately. Back then a single match prep took nine hours; after the pipeline was standing, it fell to two and a half hours. Each week I then published a model that correctly flagged set-piece overperformance. That method became my first professional credential. That habit still governs me: I start every decision with the source, the match ID, the cleaning rule — never with a guess. To me a match ID is not just a number; it is the key that tells you which delivery belonged to which innings, which bowler, under which field setting. Comparing bowling economy without clean match IDs means calling two different things by the same name. In 2026, when I took on a pressing audit for an international tournament, I learned how opponent-adjusted numbers carry more truth than raw possession. That experience taught me that a pressing audit is nothing but bookkeeping for chaos. Where each fielder stood after each ball, who ran where, which gap opened up — without that accounting, you cannot reach conclusions by reading the scorecard alone. In 2026, the empty-stadium matches taught me another lesson. That environment was a control group we never requested but received anyway. The fall in home advantage and the rise in distance covered forced me to separate venue effect from crowd effect. In regular-season cricket this discipline matters even more. In a small tournament sample one outlier can flip the whole story, but across a long league season a slowly built trend is more reliable. The betting market, however, walks the opposite road: it wants headlines, not IDs. My job is to find that gap. And to me every outlier is a question the data is asking — not an answer. I built a comparison table using eight matches each from last season and this season at Mirpur. For every match I looked at three indicators: dot-ball percentage (overs 1-15), boundary suppression rate, and a fielding pressure index — how close fielders stood per delivery to squeeze the opposition. First finding: dot-ball percentage was 42.6 percent last season and has fallen to 39.1 percent this season. That is roughly one fewer dot ball every five deliveries. On paper it looks small, but across 20 overs it amounts to about 12 fewer dot balls — directly worth eight to ten runs. Second finding: the boundary suppression rate — how often a boundary is conceded per over — is nearly unchanged. So where are the extra runs coming from? They come from the pace of singles and twos. The fielding pressure index showed that this season the average distance of infield fielders has increased by about 2.3 metres. In other words, fielders are no longer standing as close as before, so fewer dot balls are created and easy singles keep the rotation moving. My claim sits right here: the real reason runs are rising at Mirpur is not the quality of bowling but the geography of the field setting. This is a structural change, not an individual failure. I went deeper. In the ball-by-ball log I saw that the death-over yorker success rate was 38 percent last season and is 31 percent now. The curious part is that the number of yorker attempts has also risen. Bowlers are trying more and succeeding less — usually a signal of fatigue. To test fatigue I looked at the schedule. Over the last 21 days the team's pacers played matches across three different venues in three different formats within a short window. Travel, rest intervals and heat are variables I always treat as primary. Given the average temperature and humidity at Mirpur in this period, shorter death-over spells will naturally bring lower success. But I stay careful. Venue effect and crowd effect must be separated. I also added this season's attendance figures for Mirpur to the table. Average attendance is somewhat higher than last season, especially in evening matches. But the difference is not large enough to place the entire field-setting change on the crowd. So I hold two hypotheses side by side: one field-setting based, the other fatigue based. To separate them I need a test. The test is this: in matches where the team had more than two days of rest, did the fielding pressure index return to normal? In the first ten matches the answer is partly yes. In rest-rich matches the infield distance was about 0.9 metres lower — fielders stood closer again. I broke the table into two more layers. First layer: the powerplay (overs 1-6). Here dot-ball percentage barely moved — from 47 to 45 percent. Second layer: the middle overs (7-15). This is where the sharpest fall sits — from 41 to 36 percent. So the problem is not in the powerplay but in the middle overs. This matters because the betting market usually watches the powerplay and treats the middle overs as lukewarm. In betting, the edge hides in those boring columns nobody reads. Looking for why middle-over dot balls fell, I found that in the spinners' bowling charts the gap between mid-wicket and long-on had widened. The field setting had become more defensive — sacrificing singles to save the boundary. That is a conscious trade-off, and that trade-off is what is lifting the run column. Here is an example. In the middle overs, when spin is operating and long-on is kept deep, the batter easily takes a single and rotates strike. Higher strike rotation means runs rise even without boundaries. Seven or eight extra singles across 20 overs mean roughly six to eight runs — which can be the margin of defeat. I checked one more thing: the profile of opposition batters. It turned out that experienced batters spot this gap quickly, while younger batters hunt for more boundaries — which is why wickets are also falling more. This explains why runs and wickets rose together. It is not contradictory; it is the natural consequence of a field setting. I now admit the sample limitation. Eight matches each is a small sample. One rain-affected match or one exceptional pitch can flip the whole picture. So I do not treat these numbers as final proof; I treat them as a hypothesis that needs more data. My second area of interest is comparing the Indian and Bangladeshi systems. In India, a dense league schedule and rich support staff mean the same data carries a different meaning. A pacer's spell management differs there because the rest and recovery infrastructure differs. In Bangladesh, the balance between bowler workload and rest in the domestic season is the real variable. In other words, the same dot-ball number or the same pressure index does not mean the same thing in two countries. Differences in pitch, weather, travel and resources change what a metric actually means. When the definition changes, the conclusion must change too — that is my principle. Here is my caution. Correlation is not causation. The relationship between rising infield distance and rising runs is visible, but behind it could sit a captain's defensive strategy, bowler instructions, or plain sample coincidence. If my model is wrong, what evidence would force me to change my mind? The answer: a controlled situation where, at the same venue with the same bowlers, only the field setting is changed to see how much dot-ball percentage moves. I do not know whether anyone will run that. But until they do, I will make no final claim. The biggest trap in cricket analysis is using numbers to explain a person's character. When a bowler succeeds less, calling him unable to handle pressure is easy; but without process evidence that is only a story, not data. If it cannot be audited, it cannot be trusted. Another trap is clinging to an old metric past its expiry. Dot-ball percentage was a stable indicator for a long time. But when field settings, rule changes or pitch types shift, that definition must shift too. I have already fixed revision triggers in my model: a new format, a rule change, or a new data source — if any of the three occurs, I will re-verify the model. For the regular season, my advice: over the next five matches, watch dot-ball percentage and infield distance together. If dot-ball percentage returns toward 42 percent, the fatigue hypothesis strengthens. If it does not, the field-setting hypothesis holds. Not prediction, but pipeline first — that is the rule I am staying with.