Repository Public
1{"job":{"id":"b6b38e95-7371-407c-b617-67b7858a8790","tool":"event_study","params":{"symbols":["AAPL"],"event_source":"auto","max_events_per_symbol":12,"market":"SPY","model":"market_model","return_type":"simple","estimation_window":120,"gap":10,"pre":5,"post":5,"drop_overlapping":false,"confidence":0.95},"status":"done","startedAt":1795217762032,"finishedAt":1795217768275,"error":null,"progress":1,"engineVersion":"1.0.0","durationMs":6243,"projectId":null,"sessionId":null,"cancellable":false,"code":"import numpy as np, pandas as pd, statsmodels.api as sm\n# px: adjusted closes with columns \"AAPL\" and \"SPY\" on common sessions\npx = pd.read_csv(\"closes.csv\", index_col=0, parse_dates=True).dropna()\nr = px.pct_change().dropna()\nL, gap, pre, post = 120, 10, 5, 5\np0 = r.index.searchsorted(pd.Timestamp(\"2024-02-01\")) # day 0 = first session on or after the event date\nest = r.iloc[p0 - pre - gap - L: p0 - pre - gap]; evt = r.iloc[p0 - pre: p0 + post + 1]\nfit = sm.OLS(est[\"AAPL\"], sm.add_constant(est[\"SPY\"])).fit()\nar = evt[\"AAPL\"] - fit.predict(sm.add_constant(evt[\"SPY\"]))\nprint(\"CAR:\", ar.sum()) # engine: -0.052447 for AAPL 2024-02-01; repeat per event, then CAAR = mean and t = mean / (std(ddof=1) / sqrt(N))","result":{"success":true,"schema":"lab/1","engine":"event_study","engineVersion":"1.0.0","title":"Event study — AAPL, 12 earnings announcements","headline":"12 earnings announcements of AAPL (2024-02-01 → 2026-11-02), market model vs SPY: CAAR [-5, +5] = +0.49% (cross-sectional t = 0.27, p = 0.794); CAAR [+0, +1] = -0.97%; 50% of CARs positive.","summary":[{"label":"Events analysed","value":12,"unit":"count","kind":"observed","note":null},{"label":"CAAR [-5, +5]","value":0.004883055719798011,"unit":"percent","kind":"calculated","decimals":2},{"label":"CAAR [+0, +1]","value":-0.009732890725075017,"unit":"percent","kind":"calculated","decimals":2},{"label":"p-value, cross-sectional t","value":0.7936758867142268,"unit":"ratio","kind":"calculated","decimals":4},{"label":"p-value, BMP","value":0.664252215286592,"unit":"ratio","kind":"calculated","decimals":4},{"label":"p-value, Corrado rank","value":0.9251368506025343,"unit":"ratio","kind":"calculated","decimals":4},{"label":"Mean BHAR","value":0.004564488010154309,"unit":"percent","kind":"calculated","decimals":2},{"label":"Share of positive CARs","value":0.5,"unit":"percent","kind":"observed","decimals":0,"note":"expected under no effect: 49%"},{"label":"Estimation window","value":120,"unit":"days","kind":"assumption"},{"label":"Event window","value":11,"unit":"days","kind":"assumption","note":"-5 to +5, gap 10"}],"tables":[{"id":"tests","title":"Average abnormal returns and tests (12 events)","columns":[{"key":"window","label":"Window","type":"text"},{"key":"caar","label":"CAAR","unit":"percent","decimals":2,"type":"number"},{"key":"median_car","label":"Median CAR","unit":"percent","decimals":2,"type":"number"},{"key":"share_positive","label":"CAR > 0","unit":"percent","decimals":0,"type":"number"},{"key":"t_cs","label":"Cross-sectional t","decimals":2,"type":"number"},{"key":"p_cs","label":"p","decimals":4,"type":"number"},{"key":"z_patell","label":"Patell Z","decimals":2,"type":"number"},{"key":"p_patell","label":"p","decimals":4,"type":"number"},{"key":"t_bmp","label":"BMP t","decimals":2,"type":"number"},{"key":"p_bmp","label":"p","decimals":4,"type":"number"},{"key":"z_sign","label":"Generalised sign Z","decimals":2,"type":"number"},{"key":"p_sign","label":"p","decimals":4,"type":"number"},{"key":"z_rank","label":"Corrado rank Z","decimals":2,"type":"number"},{"key":"p_rank","label":"p","decimals":4,"type":"number"},{"key":"abhar","label":"Mean BHAR","unit":"percent","decimals":2,"type":"number"},{"key":"t_bhar","label":"BHAR t","decimals":2,"type":"number"}],"rows":[{"window":"[-5, +5]","caar":0.004883055719798011,"median_car":0.01022118570749719,"share_positive":0.5,"t_cs":0.26797653190113524,"p_cs":0.7936758867142268,"z_patell":0.6898439060010343,"p_patell":0.49029235442415464,"t_bmp":0.4459948416732575,"p_bmp":0.664252215286592,"z_sign":0.05292995223935906,"p_sign":0.9577877194029165,"z_rank":-0.09396513726032006,"p_rank":0.9251368506025343,"abhar":0.004564488010154309,"t_bhar":0.24475047230296462},{"window":"[-1, +1]","caar":-0.01139846945157331,"median_car":-0.007907527443972946,"share_positive":0.4166666666666667,"t_cs":-0.9153472456378164,"p_cs":0.3796381349357334,"z_patell":-1.1074578870755512,"p_patell":0.2680960102367692,"t_bmp":-0.5863246327934263,"p_bmp":0.5694934277855787,"z_sign":-0.5244877085536481,"p_sign":0.5999393666833199,"z_rank":-1.2770608757438553,"p_rank":0.20158075613592918,"abhar":-0.011054502963796545,"t_bhar":-0.8927478640727075},{"window":"[+0]","caar":-0.0021838245690399545,"median_car":-0.0015375925103555802,"share_positive":0.3333333333333333,"t_cs":-0.56785862939661,"p_cs":0.581536326948083,"z_patell":-0.33327770320231137,"p_patell":0.7389246685096096,"t_bmp":-0.33788528589679173,"p_bmp":0.7418099983775517,"z_sign":-1.1019053693466554,"p_sign":0.27050281405371734,"z_rank":-0.41046203897568784,"p_rank":0.681467044810676,"abhar":-0.002183824569039964,"t_bhar":-0.5678586293966127},{"window":"[+0, +1]","caar":-0.009732890725075017,"median_car":-0.008452377339704475,"share_positive":0.4166666666666667,"t_cs":-0.7523702177203638,"p_cs":0.46762339805338216,"z_patell":-1.0901501728100502,"p_patell":0.27564699812503046,"t_bmp":-0.4545251411696305,"p_bmp":0.6582893036336276,"z_sign":-0.5244877085536481,"p_sign":0.5999393666833199,"z_rank":-1.0803396060565282,"p_rank":0.2799909789571522,"abhar":-0.0095416547304328,"t_bhar":-0.7378075658192521},{"window":"[-5, -1]","caar":0.005769301362744531,"median_car":-0.0009830788090073304,"share_positive":0.5,"t_cs":0.6209371791604618,"p_cs":0.5472916378365567,"z_patell":0.7633689357693976,"p_patell":0.4452434006133619,"t_bmp":0.5948297577081021,"p_bmp":0.5639926534931157,"z_sign":0.05292995223935906,"p_sign":0.9577877194029165,"z_rank":-0.15297017022213413,"p_rank":0.8784217991117247,"abhar":0.005914235100061995,"t_bhar":0.635138460483656},{"window":"[+2, +5]","caar":0.008846645082128495,"median_car":0.002446983822253811,"share_positive":0.5,"t_cs":0.796758903552933,"p_cs":0.4424486777282702,"z_patell":1.0613568636189668,"p_patell":0.28852775268378605,"t_bmp":0.7671340033756796,"p_bmp":0.45915094149161173,"z_sign":0.05292995223935906,"p_sign":0.9577877194029165,"z_rank":0.7791177591668148,"p_rank":0.4359103496505946,"abhar":0.008879209290974549,"t_bhar":0.7884723424315144}],"note":"Two-sided p-values. Several windows are tested: with five windows and five tests, some nominally significant results are expected by chance."},{"id":"daily","title":"Average abnormal return by event day","columns":[{"key":"day","label":"Day","type":"integer"},{"key":"aar","label":"AAR","unit":"percent","decimals":3,"type":"number"},{"key":"t","label":"t","decimals":2,"type":"number"},{"key":"caar","label":"CAAR from -5","unit":"percent","decimals":3,"type":"number"},{"key":"caar_low","label":"Band low","unit":"percent","decimals":3,"type":"number"},{"key":"caar_high","label":"Band high","unit":"percent","decimals":3,"type":"number"},{"key":"share_positive","label":"AR > 0","unit":"percent","decimals":0,"type":"number"}],"rows":[{"day":-5,"aar":-0.002632257519531007,"t":-1.4861520021485901,"caar":-0.002632257519531007,"caar_low":-0.006530620358360002,"caar_high":0.0012661053192979882,"share_positive":0.25},{"day":-4,"aar":-0.0007437108927222355,"t":-0.20765988937396113,"caar":-0.003375968412253242,"caar_low":-0.010504135015108794,"caar_high":0.003752198190602309,"share_positive":0.3333333333333333},{"day":-3,"aar":0.006926295987228956,"t":1.4124278546598825,"caar":0.003550327574975714,"caar_low":-0.01083027555318326,"caar_high":0.017930930703134687,"share_positive":0.75},{"day":-2,"aar":0.0038845525142671092,"t":1.0988140232318155,"caar":0.007434880089242825,"caar_low":-0.010801674784731739,"caar_high":0.02567143496321739,"share_positive":0.5},{"day":-1,"aar":-0.0016655787264982923,"t":-0.8643786320748574,"caar":0.005769301362744533,"caar_low":-0.01468066863368201,"caar_high":0.026219271359171077,"share_positive":0.3333333333333333},{"day":0,"aar":-0.002183824569039955,"t":-0.5678586293966101,"caar":0.0035854767937045767,"caar_low":-0.016250543004599154,"caar_high":0.023421496592008303,"share_positive":0.3333333333333333},{"day":1,"aar":-0.007549066156035063,"t":-0.7362894638347846,"caar":-0.003963589362330487,"caar_low":-0.03516041825858465,"caar_high":0.027233239533923682,"share_positive":0.3333333333333333},{"day":2,"aar":-0.004588133276788595,"t":-0.6525205001799469,"caar":-0.00855172263911908,"caar_low":-0.04019674014048964,"caar_high":0.02309329486225148,"share_positive":0.25},{"day":3,"aar":0.00459361323563581,"t":1.4450513499436655,"caar":-0.003958109403483271,"caar_low":-0.03718902277533365,"caar_high":0.02927280396836711,"share_positive":0.75},{"day":4,"aar":0.0022703237372825526,"t":0.39109763627333444,"caar":-0.0016877856662007177,"caar_low":-0.04009857248821753,"caar_high":0.036723001155816094,"share_positive":0.4166666666666667},{"day":5,"aar":0.006570841385998728,"t":2.0425223728059927,"caar":0.004883055719798011,"caar_low":-0.03522319201364572,"caar_high":0.04498930345324174,"share_positive":0.75}],"note":"t and band from the cross-section of events."},{"id":"events","title":"Events","columns":[{"key":"symbol","label":"Symbol","type":"text"},{"key":"date","label":"Event date","type":"date"},{"key":"day0","label":"Day 0","type":"date"},{"key":"car","label":"CAR [-5, +5]","unit":"percent","decimals":2,"type":"number"},{"key":"t_car","label":"t","decimals":2,"type":"number"},{"key":"p_car","label":"p","decimals":4,"type":"number"},{"key":"bhar","label":"BHAR","unit":"percent","decimals":2,"type":"number"},{"key":"ar0","label":"AR day 0","unit":"percent","decimals":2,"type":"number"},{"key":"alpha","label":"Alpha (daily)","decimals":5,"type":"number"},{"key":"beta","label":"Beta","decimals":3,"type":"number"},{"key":"sigma","label":"Residual s.d.","unit":"percent","decimals":2,"type":"number"},{"key":"eps_surprise","label":"EPS surprise","unit":"percent","decimals":1,"type":"number"},{"key":"zero_returns","label":"Zero-return days (estimation)","unit":"percent","decimals":0,"type":"number"},{"key":"contaminated","label":"Other event in estimation window","type":"text"}],"rows":[{"symbol":"AAPL","date":"2024-02-01","day0":"2024-02-01","alpha":-0.0008095671533371938,"beta":1.1141646545075297,"sigma":0.009509513655010153,"car":-0.05244656050035842,"t_car":-1.6480533381777969,"p_car":0.1020019457925538,"bhar":-0.052336069522264106,"ar0":-0.00042465100777872543,"eps_surprise":0.038095238095238126,"zero_returns":0,"contaminated":false},{"symbol":"AAPL","date":"2024-05-02","day0":"2024-05-02","alpha":-0.0017315770481574838,"beta":0.9374864546833235,"sigma":0.009574018621725437,"car":0.08227533186488104,"t_car":2.567592265915239,"p_car":0.011489736573202325,"bhar":0.08396801619758643,"ar0":0.014944539337706004,"eps_surprise":0.013245033112582794,"zero_returns":0,"contaminated":true},{"symbol":"AAPL","date":"2024-08-01","day0":"2024-08-01","alpha":0.0005653369430253501,"beta":1.0591344766229431,"sigma":0.013974281505791428,"car":-0.009075051343449006,"t_car":-0.1908354690469918,"p_car":0.8489824484894913,"bhar":-0.009249977615699456,"ar0":-0.002354924832736336,"eps_surprise":0.04477611940298495,"zero_returns":0,"contaminated":true},{"symbol":"AAPL","date":"2024-10-31","day0":"2024-10-31","alpha":0.0013531064752691701,"beta":1.1205789275674005,"sigma":0.013112719454676486,"car":-0.06288531109643411,"t_car":-1.430716056901878,"p_car":0.15515447945824432,"bhar":-0.06381487269338815,"ar0":0.002406856798986466,"eps_surprise":0.024999999999999883,"zero_returns":0,"contaminated":true},{"symbol":"AAPL","date":"2025-01-30","day0":"2025-01-30","alpha":1.7419699797444023e-06,"beta":0.8474750782645085,"sigma":0.010195743380916705,"car":0.04303173882171935,"t_car":1.2643650963950548,"p_car":0.2085907052823686,"bhar":0.04204632772140393,"ar0":-0.01196535663564883,"eps_surprise":0.0169491525423729,"zero_returns":0.008333333333333333,"contaminated":true},{"symbol":"AAPL","date":"2025-05-01","day0":"2025-05-01","alpha":-0.0008787756329916249,"beta":1.1046845119459134,"sigma":0.014052548306539174,"car":-0.08422745122868355,"t_car":-1.7939235142466834,"p_car":0.07538626539927722,"bhar":-0.08586477216717991,"ar0":-0.0031002086802598486,"eps_surprise":0.012269938650306761,"zero_returns":0.008333333333333333,"contaminated":true},{"symbol":"AAPL","date":"2025-07-31","day0":"2025-07-31","alpha":-0.0014743424966472557,"beta":1.311710685334176,"sigma":0.014985226380772591,"car":0.04920783772309262,"t_car":0.9850861140406699,"p_car":0.3265973248349887,"bhar":0.047679983569948625,"ar0":-0.0007202601879748244,"eps_surprise":0.09027777777777786,"zero_returns":0,"contaminated":true},{"symbol":"AAPL","date":"2025-10-30","day0":"2025-10-30","alpha":2.8640253929759947e-05,"beta":1.1611479334186152,"sigma":0.013178107768425687,"car":0.038217506961568545,"t_car":0.8669130423833419,"p_car":0.3877490770890211,"bhar":0.03918611519406334,"ar0":0.019074706862792924,"eps_surprise":0.06936416184971105,"zero_returns":0,"contaminated":true},{"symbol":"AAPL","date":"2026-01-29","day0":"2026-01-29","alpha":0.001105649462610019,"beta":0.9208532297059573,"sigma":0.012227877820465444,"car":0.10793628663691933,"t_car":2.642395460693964,"p_car":0.009348864951928873,"bhar":0.11243633965116029,"ar0":0.007883683543071639,"eps_surprise":0.06741573033707872,"zero_returns":0,"contaminated":true},{"symbol":"AAPL","date":"2026-04-30","day0":"2026-04-30","alpha":0.00025235860907742966,"beta":0.8682261441660205,"sigma":0.011458674279116638,"car":0.024593608208550596,"t_car":0.6428414403875192,"p_car":0.5215741558730091,"bhar":0.02452930447379975,"ar0":-0.004693206454396138,"eps_surprise":0.030769230769230684,"zero_returns":0,"contaminated":true},{"symbol":"AAPL","date":"2026-07-30","day0":"2026-07-30","alpha":0.0011238739734194818,"beta":0.82061663680356,"sigma":0.015438458352684948,"car":-0.07388003061667403,"t_car":-1.4288099174088071,"p_car":0.15570067612940253,"bhar":-0.07731402753743466,"ar0":-0.028950819087054793,"eps_surprise":0.06878306878306885,"zero_returns":0,"contaminated":true},{"symbol":"AAPL","date":"2026-11-02","day0":"2026-11-02","alpha":0.0016175810308080483,"beta":0.3412517509317049,"sigma":0.017103117807796362,"car":-0.004151236793556218,"t_car":-0.0713662324970156,"p_car":0.9432270996229405,"bhar":-0.0064925111501443755,"ar0":-0.018306254485186996,"eps_surprise":0.030150753768844137,"zero_returns":0,"contaminated":true}],"note":"Per-event t = CAR / (σ̂ √Σ(1 + prediction-error correction)), Student with the estimation degrees of freedom; it assumes normal, serially uncorrelated abnormal returns with unchanged variance in the event window."},{"id":"by_surprise","title":"CAAR by sign of the earnings surprise","columns":[{"key":"group","label":"Group","type":"text"},{"key":"n","label":"Events","type":"integer"},{"key":"caar","label":"CAAR [-5, +5]","unit":"percent","decimals":2,"type":"number"},{"key":"t_cs","label":"t","decimals":2,"type":"number"},{"key":"p_cs","label":"p","decimals":4,"type":"number"},{"key":"caar_01","label":"CAAR [+0, +1]","unit":"percent","decimals":2,"type":"number"},{"key":"t_01","label":"t","decimals":2,"type":"number"}],"rows":[{"group":"Positive EPS surprise","n":12,"caar":0.004883055719798011,"t_cs":0.26797653190113524,"p_cs":0.7936758867142268,"caar_01":-0.009732890725075017,"t_01":-0.7523702177203638}],"note":"EPS surprise = (actual − consensus estimate) / |estimate| as reported by FMP. Splitting on the surprise conditions on information released at the event: it describes the reaction, it is not a trading rule."}],"charts":[{"id":"caar","title":"Cumulative average abnormal return","type":"line","x":{"label":"Event day","type":"number","unit":"days"},"y":{"label":"CAAR","unit":"percent"},"series":[{"name":"CAAR","kind":"calculated","x":[-5,-4,-3,-2,-1,0,1,2,3,4,5],"y":[-0.002632257519531007,-0.003375968412253242,0.003550327574975714,0.007434880089242825,0.005769301362744533,0.0035854767937045767,-0.003963589362330487,-0.00855172263911908,-0.003958109403483271,-0.0016877856662007177,0.004883055719798011],"lower":[-0.006530620358360002,-0.010504135015108794,-0.01083027555318326,-0.010801674784731739,-0.01468066863368201,-0.016250543004599154,-0.03516041825858465,-0.04019674014048964,-0.03718902277533365,-0.04009857248821753,-0.03522319201364572],"upper":[0.0012661053192979882,0.003752198190602309,0.017930930703134687,0.02567143496321739,0.026219271359171077,0.023421496592008303,0.027233239533923682,0.02309329486225148,0.02927280396836711,0.036723001155816094,0.04498930345324174],"style":"band","line":true}],"annotations":[{"x":0,"text":"event"}],"note":"Cumulated from day -5. Band: 95% pointwise interval from the cross-section of events."},{"id":"aar","title":"Average abnormal return by day","type":"bar","x":{"label":"Event day","type":"number","unit":"days"},"y":{"label":"AAR","unit":"percent"},"series":[{"name":"AAR","kind":"calculated","x":[-5,-4,-3,-2,-1,0,1,2,3,4,5],"y":[-0.002632257519531007,-0.0007437108927222355,0.006926295987228956,0.0038845525142671092,-0.0016655787264982923,-0.002183824569039955,-0.007549066156035063,-0.004588133276788595,0.00459361323563581,0.0022703237372825526,0.006570841385998728],"lower":[-0.006530620358360002,-0.008626295455465428,-0.0038669453326713567,-0.0038964206592931446,-0.005906676231636056,-0.010648194431320243,-0.03011543902716923,-0.02006414246262944,-0.002403006340013944,-0.010506406140139578,-0.0005097784262733914],"upper":[0.0012661053192979882,0.007138873670020957,0.01771953730712927,0.011665525687827364,0.002575518778639472,0.006280545293240333,0.015017306715099103,0.010887875909052248,0.011590232811285565,0.015047053614704682,0.013651461198270848],"style":"bar"}]},{"id":"cars","title":"CAR [-5, +5] per event","type":"bar","x":{"label":"Event","type":"category"},"y":{"label":"CAR","unit":"percent"},"series":[{"name":"CAR","kind":"calculated","x":["AAPL 2024-02-01","AAPL 2024-05-02","AAPL 2024-08-01","AAPL 2024-10-31","AAPL 2025-01-30","AAPL 2025-05-01","AAPL 2025-07-31","AAPL 2025-10-30","AAPL 2026-01-29","AAPL 2026-04-30","AAPL 2026-07-30","AAPL 2026-11-02"],"y":[-0.05244656050035842,0.08227533186488104,-0.009075051343449006,-0.06288531109643411,0.04303173882171935,-0.08422745122868355,0.04920783772309262,0.038217506961568545,0.10793628663691933,0.024593608208550596,-0.07388003061667403,-0.004151236793556218],"style":"bar"}],"note":"Chronological order."},{"id":"car_distribution","title":"Distribution of CARs across events","type":"histogram","x":{"label":"CAR [-5, +5]","type":"number","unit":"percent"},"y":{"label":"Events","unit":"count"},"series":[{"name":"Events","kind":"calculated","x":[-0.06821380640654998,-0.036186516762282836,-0.004159227118015689,0.027868062526251458,0.05989535217051861,0.09192264181478577],"y":[4,0,2,3,1,2],"style":"bar"}]}],"diagnostics":[{"name":"Cross-sectional t-test of CAAR","statistic":0.26797653190113524,"pvalue":0.7936758867142268,"conclusion":"Fails to reject a zero average abnormal return over [-5, +5] at the 5% level (which does not show there is no effect).","passed":null,"df":11},{"name":"Boehmer-Musumeci-Poulsen standardised cross-sectional test","statistic":0.4459948416732575,"pvalue":0.664252215286592,"conclusion":"Fails to reject a zero average abnormal return over [-5, +5] at the 5% level (which does not show there is no effect).","passed":null,"df":11},{"name":"Patell standardised residual test","statistic":0.6898439060010343,"pvalue":0.49029235442415464,"conclusion":"Fails to reject a zero average abnormal return over [-5, +5] at the 5% level (which does not show there is no effect).","passed":null},{"name":"Generalised sign test (Cowan)","statistic":0.05292995223935906,"pvalue":0.9577877194029165,"conclusion":"Fails to reject a zero average abnormal return over [-5, +5] at the 5% level (which does not show there is no effect).","passed":null},{"name":"Corrado rank test","statistic":-0.09396513726032006,"pvalue":0.9251368506025343,"conclusion":"Fails to reject a zero average abnormal return over [-5, +5] at the 5% level (which does not show there is no effect).","passed":null},{"name":"Agreement of the tests","statistic":0,"pvalue":null,"conclusion":"0 of 5 tests reject at the 5% level. They agree.","passed":true},{"name":"Estimation windows free of other events","statistic":11,"pvalue":null,"conclusion":"11 of 12 estimation windows contain the event window of another event of the same symbol (unavoidable for quarterly announcements when the estimation window exceeds about 50 sessions). Those days inflate the residual variance, which makes the standardised tests conservative; alpha and beta can also be affected.","passed":false},{"name":"Event windows separated in calendar time","statistic":0,"pvalue":null,"conclusion":"No two events of different symbols share calendar days in their event windows.","passed":true},{"name":"Thin trading (zero-return days in estimation windows)","statistic":0.008333333333333333,"pvalue":null,"conclusion":"No estimation window has more than 20% of zero-return days.","passed":true},{"name":"Normality of estimation-period abnormal returns (Jarque-Bera, pooled)","statistic":526.5768906271107,"pvalue":4.521484552599604e-115,"conclusion":"Rejects normality at the 5% level (fat tails are the rule for daily returns): prefer the BMP and rank tests to exact small-sample t interpretations.","passed":false,"df":2}],"assumptions":["Normal returns: market model R = α + β·R(SPY) + ε, estimated by OLS on the 120 sessions ending 11 sessions before the event window; simple returns on adjusted closes.","Day 0 is the first session on or after the event date. Announcements made after the close move the price on day +1: read the [0, +1] or [-1, +1] window rather than day 0 alone.","Without the event, the model's parameters and the residual variance would have stayed what they were in the estimation window.","The tests treat events as independent draws; the parametric ones also assume approximately normal abnormal returns.","Earnings dates are FMP's announcement dates (announced results only); the time of day is not available. The average mixes good and bad news, so a CAAR near zero can hide large reactions of both signs (see the split by surprise and the distribution of CARs).","An abnormal return is abnormal relative to the chosen benchmark only; it measures association with the event date, and other news on the same days is not separated out."],"warnings":["11 of 12 estimation windows contain another event of the same symbol (see diagnostics). Shorten the estimation window or accept conservative tests."],"methodology":"**Abnormal returns.** $AR_{it} = R_{it} - E[R_{it}]$ with $E[R_{it}] = \\hat\\alpha_i + \\hat\\beta_i R_{mt}$ (market model), $R_{mt}$ (market-adjusted) or $\\bar R_i$ (mean-adjusted), parameters from the estimation window of $L$ sessions. $CAR_i(\\tau_1,\\tau_2) = \\sum_{t=\\tau_1}^{\\tau_2}AR_{it}$, $BHAR_i = \\prod(1+R_{it}) - \\prod(1+E[R_{it}])$, $CAAR = \\frac1N\\sum_i CAR_i$.\n\n**Cross-sectional t.** $t = CAAR / (s_{CAR}/\\sqrt N)$, $s_{CAR}$ the standard deviation of the $CAR_i$; robust to event-induced variance, not to cross-correlation.\n\n**Patell (1976).** $SAR_{it} = AR_{it}/(\\hat\\sigma_i\\sqrt{C_{it}})$ with $C_{it} = 1 + \\frac1L + \\frac{(R_{mt}-\\bar R_m)^2}{\\sum_\\tau(R_{m\\tau}-\\bar R_m)^2}$ (market model; $1+\\frac1L$ mean-adjusted; $1$ market-adjusted); $Z = \\sum_i \\frac{1}{\\sqrt m}\\sum_t SAR_{it}\\Big/\\sqrt{\\sum_i \\frac{d_i}{d_i-2}}$, $m$ the window length, $d_i$ the estimation degrees of freedom. Assumes no event-induced variance and independence.\n\n**BMP (1991).** $SCAR_i = \\frac{1}{\\sqrt m}\\sum_t SAR_{it}$; $t = \\overline{SCAR}/(s_{SCAR}/\\sqrt N)$: standardisation plus a cross-sectional variance, robust to event-induced variance. The correlation between the prediction errors of different days of one event is ignored.\n\n**Generalised sign (Cowan 1992).** $Z = (w - N\\hat p)/\\sqrt{N\\hat p(1-\\hat p)}$, $w$ the number of positive CARs, $\\hat p$ the share of positive abnormal returns in the estimation windows.\n\n**Rank (Corrado 1989).** Ranks $K_{it}$ of $AR_{it}$ over estimation and event window together ($T=L+m_{w}$ days), $U_{it} = K_{it}/(T+1) - \\tfrac12$, $\\bar U_t = \\frac1N\\sum_i U_{it}$, $Z = \\sum_{t=\\tau_1}^{\\tau_2}\\bar U_t\\Big/\\big(\\sqrt m\\, S_U\\big)$, $S_U^2 = \\frac1T\\sum_t \\bar U_t^2$.\n\n**Bands.** CAAR path: $\\pm t_{N-1}\\, s_{CAR(\\tau)}/\\sqrt N$ pointwise.","provenance":{"data":[{"symbol":"SPY","provider":"FMP","retrievedAt":"2026-11-20T23:36:06+00:00","start":"2023-06-23","end":"2026-11-20","observations":858,"frequency":"1d","adjusted":"split_div","currency":null},{"symbol":"AAPL","provider":"FMP","retrievedAt":"2026-11-20T23:36:07+00:00","start":"2023-06-23","end":"2026-11-20","observations":858,"frequency":"1d","adjusted":"split_div","currency":null}],"spec":{"symbols":["AAPL"],"events":null,"event_source":"auto","max_events_per_symbol":12,"market":"SPY","model":"market_model","return_type":"simple","estimation_window":120,"gap":10,"pre":5,"post":5,"drop_overlapping":false,"confidence":0.95,"event_source_used":"earnings","events_used":[{"symbol":"AAPL","date":"2024-02-01","day0":"2024-02-01"},{"symbol":"AAPL","date":"2024-05-02","day0":"2024-05-02"},{"symbol":"AAPL","date":"2024-08-01","day0":"2024-08-01"},{"symbol":"AAPL","date":"2024-10-31","day0":"2024-10-31"},{"symbol":"AAPL","date":"2025-01-30","day0":"2025-01-30"},{"symbol":"AAPL","date":"2025-05-01","day0":"2025-05-01"},{"symbol":"AAPL","date":"2025-07-31","day0":"2025-07-31"},{"symbol":"AAPL","date":"2025-10-30","day0":"2025-10-30"},{"symbol":"AAPL","date":"2026-01-29","day0":"2026-01-29"},{"symbol":"AAPL","date":"2026-04-30","day0":"2026-04-30"},{"symbol":"AAPL","date":"2026-07-30","day0":"2026-07-30"},{"symbol":"AAPL","date":"2026-11-02","day0":"2026-11-02"}]},"seed":null,"software":{"python":"3.14.4","numpy":"2.4.6","pandas":"3.0.3","scipy":"1.17.1","statsmodels":"0.14.6","arch":"8.0.0","scikit-learn":"1.8.0"},"computedAt":"2026-11-20T23:36:07+00:00","durationMs":5123,"reproduce":"import numpy as np, pandas as pd, statsmodels.api as sm\n# px: adjusted closes with columns \"AAPL\" and \"SPY\" on common sessions\npx = pd.read_csv(\"closes.csv\", index_col=0, parse_dates=True).dropna()\nr = px.pct_change().dropna()\nL, gap, pre, post = 120, 10, 5, 5\np0 = r.index.searchsorted(pd.Timestamp(\"2024-02-01\")) # day 0 = first session on or after the event date\nest = r.iloc[p0 - pre - gap - L: p0 - pre - gap]; evt = r.iloc[p0 - pre: p0 + post + 1]\nfit = sm.OLS(est[\"AAPL\"], sm.add_constant(est[\"SPY\"])).fit()\nar = evt[\"AAPL\"] - fit.predict(sm.add_constant(evt[\"SPY\"]))\nprint(\"CAR:\", ar.sum()) # engine: -0.052447 for AAPL 2024-02-01; repeat per event, then CAAR = mean and t = mean / (std(ddof=1) / sqrt(N))"}}}}2