HomeAsian CricketThe Mirpur Ledger: How an Expected-Runs Model Taught the BPL Scoreboard to Be Read Twice
Asian Cricket

The Mirpur Ledger: How an Expected-Runs Model Taught the BPL Scoreboard to Be Read Twice

**মূল উত্তর:** বিপিএলের শেষ তিন আসরের ১৩২ ম্যাচ ও ৩০,৪১২ ডেলিভারির এক্সপেক্টেড-রান মডেল দেখায়, বাংলাদেশের ফ্র্যাঞ্চাইজি Batting তিন ধাপেই পার-এর নিচে ছিল; ঘাটতি মূলত ডট-বলের খরচে। **মূল তথ্য:** - পাওয়ারপ্লে রান-রেট ৭.৪, পার ৮.১; ডট-শতাংশ ৪৬ বনাম পার ৩৮। - ডট-প্রেশার ইনডেক্সে শীর্ষ চার দল (৩.১, ২.৮, ২.৬, ২.৪) সবাই প্লে-অফে খেলেছে। - মিডল ওভারে মেহেদী হাসান মিরাজের Economy ৬.২, সেই ধাপের পার-Economy ৭.৬। - ডেথ ওভারে মুস্তাফিজুর রহমানের Economy ৮.৯ এবং তাসকিন আহমেদের ৯.৬, পার ১০.৪। - Innings পার-স্কোরের সাথে ±৪.৭ রানের ত্রুটি-বালতি প্রকাশ করা হয়েছে। **সূত্র:** লেখকের এক্সপেক্টেড-রান লেজার, বিপিএল ২০২৪-২৫ ডেটাসেট; বিশ্লেষণ প্রকাশিত ১২ ফেব্রুয়ারি, ২০২৬। | Cross-checked: cricsultan.com **সম্পর্কিত প্রশ্নোত্তর:** প্রশ্ন: ডট-প্রেশার ইনডেক্স কী মাপে? উত্তর: প্রেশার ডেলিভারির সংখ্যাকে উইকেট-বল হার দিয়ে ভাগ করে Bowling আক্রমণ ব্যাটারকে কতটা অচল করে, তা মাপে। প্রশ্ন: ফ্র্যাঞ্চাইজি নিলামে পার-ডিফারেনশিয়াল কেন গুরুত্বপূর্ণ? উত্তর: কারণ স্ট্রাইক-রেটের নিচে লুকানো ডট-বল-খরচই মূল্য নির্ধারণ করে, যা cricsultan.com Player Depth Index-এর সাথে মিলিয়ে যাচাই করা যায়। প্রশ্ন: মডেলের প্রধান সীমাবদ্ধতা কী? উত্তর: ফিল্ডিং ত্রুটি, আম্পায়ারের ধরন ও শিশিরের প্রভাব মডেলের ভেরিয়েবলে ধরা পড়ে না, তাই ±৪.৭ রানের ত্রুটি-বালতি প্রকাশ করা হয়।

The Mirpur Ledger: How an Expected-Runs Model Taught the BPL Scoreboard to Be Read Twice At the Sher-e-Bangla Stadium, the crowd on the eastern side had stood up. Eleven runs needed off five balls, six wickets in hand. When the sixteenth over ended, the number glowing on my screen read 0.38 — the chasing side's win probability. The broadcast graphic meanwhile said the game had come alive; the scoreboard said the road was easy. Two truths cannot both be true at once. The batter who finished that night on 58 off 47 and took the man-of-the-match award had a par score of 64.3. Adjusting for wickets lost, ball age and the spin matchups he faced, that innings sat 6.3 runs below par. Protecting his wicket was not irrational, but the victory came from somewhere else — the yorker line in the final over, and a diving stop at deep midwicket two balls earlier by Tanzim Hasan Sakib. I do not chase results; I audit the process until it confesses. I built the first xG ledger in Sylhet, and it taught me one principle — the scoreboard first, then the process that produced it, then the uncertainty around both. Moving from expected goals in football to cricket, I found that runs per ball can be measured the same way; only the phases differ, and only the weight of a dot ball differs. Every ball in cricket carries a potential value, and that value depends on four things — match phase, pitch behaviour, bowler type and batter handedness. When a pacer bowled with the new ball in the powerplay, I logged a pressure-deliveries-per-dot ratio (PDD) of 2.1. If the same bowler returns in the twenty-fourth over and concedes at 9.4, the number is meaningless, because the context has moved. Across the last three BPL seasons I logged 30,412 deliveries from 132 matches, training two junior writers in coordinate logging. Alongside each delivery I recorded pitch age, dew probability, ball type, batter's crease position and field placement. That base produced an expected-runs table published with 90 percent confidence intervals attached. I do not hide the model's limits. Slow-motion dropped catches, umpiring tendencies, home-venue familiarity — none of these are variables. So next to every innings par score I print an error bucket of plus or minus 4.7 runs. That habit keeps me away from two traps at once: scoreboard worship and number worship. In the powerplay, a stable picture has formed for Bangladesh. Over the last ten franchise matches, the average powerplay run rate was 7.4, while par on the same pitches was 8.1. The entire 0.7-run shortfall came from dot balls — powerplay dot percentage of 46 against a par of 38. The problem was not a shortage of scoring shots; it was a failure to locate the boundary ball. | Phase | Bangladesh run rate | Par run rate | Gap | Dot % | | Powerplay (1-6) | 7.4 | 8.1 | -0.7 | 46 | | Middle (7-15) | 7.1 | 7.6 | -0.5 | 39 | | Death (16-20) | 9.8 | 10.4 | -0.6 | 34 | | Full innings | 7.9 | 8.5 | -0.6 | 41 | Reading the table, one thing is clear — the deficit is not concentrated in any single phase, it is spread across all three. In franchise cricket that is dangerous, because each phase has a different coach, a different plan and a different budget. Anyone who believes the problem sits only in the powerplay will invest in the wrong place. Middle overs are a story about spin. Mehidy Hasan Miraz kept an economy of 6.2 last season, when par economy from the seventh to the fifteenth over was 7.6. On average, 1.9 balls per over landed on a length from which the batter could only defend. In the same phase, Rishad Hossain's wicket differential was plus 3.8 — he took 3.8 more wickets than par — while his economy ran 0.9 above par. At a franchise auction these two spinners cost roughly the same, yet their jobs are entirely different. Out of this comes an index I built — the Dot-Pressure Index (DPI). In football, PPDA shows how high a team presses; in cricket, DPI shows how thoroughly a bowling attack immobilises the batter. The calculation is simple: pressure deliveries divided by wickets lost per ball. Last season, the four best DPI figures in the BPL were 3.1, 2.8, 2.6 and 2.4 — and all four sides reached the playoffs. The bottom three sides recorded 1.2, 1.1 and 0.9. At the death, the pairing of Mustafizur Rahman and Taskin Ahmed is exceptional in par terms. Last season Mustafizur's death economy was 8.9 against a par of 10.4; Taskin's was 9.6 against the same par. Both bowled below par even though their wicket percentages were nearly identical — 12.1 percent against 11.8. In practical terms, they did not break the opposition's expected 9-10 runs by magic; they broke it through continuity of plan. One yorker in the right place does not change a scoreboard; five in a row do. At batter level I found two opposite portraits. Litton Das's boundary percentage runs above par, but his dot percentage sits seven points above par. Towhid Hridoy's boundary percentage is lower than Litton's, yet his par differential is plus 9.4 runs, because he holds dot balls at 31 percent. In T20 cricket, strike rate is a product, but the dot-ball cost hidden beneath strike rate is the real price. The franchise auction is not a bazaar; it is a probability engine with agents. Last season one middle-order batter was priced on a strike rate of 138, with a par differential of minus 6.2. Another went near the bottom of the table on a strike rate of 129, with a par differential of plus 11.7. By mid-season the second had become his team's mainstay. Had scouting read par differential instead of strike rate, two teams would have saved at least two matches. A spreadsheet is a monastery, and I take vows in columns and rows. But even monastery walls have limits. My model measures dew, pitch abrasion, a bowler's line and length — it does not measure a slip fielder arriving a second late. I keep a separate stadium-effect variable, because the air in Mirpur is not the air in Sylhet; the same delivery yields six in one venue and four in another. The World Cup final gave us two truths: the scoreboard and the process. France beat Croatia 4-2, yet expected goals stood at 2.1 against 1.8 — and Croatia's midfield control that evening, with a PPDA of 12.4, told a different story. Cricket shows the same gap. Last season a side led the par differential in four of its seven matches and still missed the playoffs — the toss, dew and two run-outs pushed them outside their own arithmetic. Correlation and causation are different things. Teams with high DPI win more matches — true, but that does not mean raising DPI produces wins. Good teams build good squads first, and good DPI follows; read the sequence backwards and you will lose eight matches. Numbers point toward process; they never replace it. Empty stadiums taught me that silence has its own expected goals. In venues without crowds, my win-probability model tightened by roughly eleven percentage points — because the noise was gone, leaving only the sound of the ball and the batter's decision. That silence has value in cricket: boundaries off pressure deliveries rose by three percentage points, because batters could trust their own rhythm. Going into the next round, I will watch two things. First, middle-over spin matchups: when an opposition sends two right-handed middle-order batters against a left-arm spinner, DPI naturally drops 0.7 points — an early signal of collapse. Second, death-over line-and-length discipline: if par economy of 10.4 is breached, I will treat that attack as incomplete no matter how many wickets it takes. The question is simple: if franchise auctions are still willing to buy strike rate, who will buy par differential? The side that answers first will sit two steps ahead in next season's table — while the rest are still hunting for a man-of-the-match innings.

The Mirpur Ledger: How an Expected-Runs Model Taught the BPL Scoreboard to Be Read Twice

The Mirpur Ledger: How an Expected-Runs Model Taught the BPL Scoreboard to Be Read Twice

Related Players