Riset tentang sistem trading yang bertahan di pasar nyata.https://marketmaker.cc/idhttps://marketmaker.cc/id/blog/xgboost-gradient-boosting-trading/https://marketmaker.cc/id/blog/xgboost-gradient-boosting-trading/Return-direction classifiers are imbalanced problems, and 0.5 is the wrong decision threshold. Comparing scale_pos_weight, focal loss, and precision-constrained threshold optimization on crypto data - plus the engineering differences between XGBoost, LightGBM, and CatBoost.Tue, 25 Aug 2026 00:00:00 GMTmachine-learningxgboostgradient-boostingquantpredictionhttps://marketmaker.cc/id/blog/transfer-entropy-crypto-information-flow/https://marketmaker.cc/id/blog/transfer-entropy-crypto-information-flow/DCC-GARCH tells you when crypto dependence tightens. Transfer entropy tells you which way it points. A directed information-flow measure, its null calibration, and an honest account of what it does and does not add over average pairwise correlation.Mon, 24 Aug 2026 00:00:00 GMTcausal-inferencetransfer-entropyinformation-theorycryptonetworkhttps://marketmaker.cc/id/blog/trade-classification-algorithms/https://marketmaker.cc/id/blog/trade-classification-algorithms/Reconstructing trade direction from OHLCV, bar history, or venues without an aggressor flag - the classic rules, the stale-quote problem, bulk volume classification, and how to measure them against crypto's free ground truth.Sun, 23 Aug 2026 00:00:00 GMTmicrostructuretrade-classificationLee-Readyorder-flowdeep-learninghttps://marketmaker.cc/id/blog/tick-data-forecasting-deep-learning/https://marketmaker.cc/id/blog/tick-data-forecasting-deep-learning/Sequence models fed tick data still assume regular spacing. Three ways to tell a Transformer when a tick actually happened - learnable-timescale continuous encoding, ODE-RNN latent state, and delta_t as a plain feature - and the ablation that decides between them.Sat, 22 Aug 2026 00:00:00 GMTHFTtick-datadeep-learninghigh-frequencypredictionhttps://marketmaker.cc/id/blog/synthetic-control-trading-evaluation/https://marketmaker.cc/id/blog/synthetic-control-trading-evaluation/This series has priced the selection route to a false edge - DSR prices the winner, PBO prices the search. Neither touches confounding: the strategy that made money because volatility doubled the week you deployed it. The Synthetic Control Method builds a weighted counterfactual from a donor pool of untouched instruments and gives you a falsification criterion and a placebo p-value.Fri, 21 Aug 2026 00:00:00 GMTcausal-inferencesynthetic-controlcounterfactualevaluationeconometricshttps://marketmaker.cc/id/blog/probabilistic-forecasting-trading/https://marketmaker.cc/id/blog/probabilistic-forecasting-trading/How to evaluate a predictive distribution honestly - CRPS as a proper scoring rule, the PIT histogram as a calibration diagnostic, and DeepAR sampling in GluonTS.Thu, 20 Aug 2026 00:00:00 GMTforecastingprobabilisticquantile-regressionDeepARuncertaintyhttps://marketmaker.cc/id/blog/pinn-options-pricing/https://marketmaker.cc/id/blog/pinn-options-pricing/Putting the Heston, free-boundary, and jump-diffusion pricing PDEs into a neural network loss - the log-price residual, the mixed-partial autograd trick, and what still needs measuring.Wed, 19 Aug 2026 00:00:00 GMTdeep-learningPINNoptionsBlack-ScholesPDEhttps://marketmaker.cc/id/blog/pcmci-causal-discovery-crypto/https://marketmaker.cc/id/blog/pcmci-causal-discovery-crypto/How PCMCI's two-stage MCI test recovers directed causal links between crypto assets where correlation and bivariate Granger cannot - the construction, the tigramite pipeline, and the real-data study it still needs.Tue, 18 Aug 2026 00:00:00 GMTcausal-inferencePCMCIcausal-discoverytime-seriescryptohttps://marketmaker.cc/id/blog/order-flow-imbalance-prediction/https://marketmaker.cc/id/blog/order-flow-imbalance-prediction/Turning raw book updates into a signed flow quantity: the CKS event decomposition, multi-level OFI with PCA reduction, and Lee-Ready trade classification - plus an honest accounting of what the headline R-squared actually measures.Mon, 17 Aug 2026 00:00:00 GMTmicrostructureorder-flowimbalancepredictionHFThttps://marketmaker.cc/id/blog/neural-ode-continuous-finance/https://marketmaker.cc/id/blog/neural-ode-continuous-finance/Neural ODEs, Neural SDEs and continuous normalizing flows for irregularly-sampled market data - and the one ablation that decides whether continuous dynamics are worth their solver cost.Sun, 16 Aug 2026 00:00:00 GMTdeep-learningneural-ODEcontinuous-timeSDEdynamicshttps://marketmaker.cc/id/blog/multi-task-learning-trading/https://marketmaker.cc/id/blog/multi-task-learning-trading/Does jointly predicting return, volume, and volatility actually help? Measuring loss-balancing schemes and diagnosing negative transfer through gradient cosine similarity - with a classical baseline and purged walk-forward folds.Sat, 15 Aug 2026 00:00:00 GMTdeep-learningmulti-taskshared-representationpredictionauxiliaryhttps://marketmaker.cc/id/blog/model-pruning-low-latency-trading/https://marketmaker.cc/id/blog/model-pruning-low-latency-trading/Magnitude and structured pruning, the Lottery Ticket Hypothesis, movement pruning, distillation and 2:4 sparsity - the methods behind shrinking a trading model, and what still has to be measured before any of it ships.Fri, 14 Aug 2026 00:00:00 GMTmodel-compressionpruninglottery-ticketlatencydeploymenthttps://marketmaker.cc/id/blog/liquidity-prediction-execution/https://marketmaker.cc/id/blog/liquidity-prediction-execution/Our VWAP article shipped a static, weekly-refit volume curve and called the forecaster the weakest link. This is the follow-up: a Bayesian intraday updater run against the same 500-parent BTCUSDT harness, with the IS delta conditioned on realized curve error.Thu, 13 Aug 2026 00:00:00 GMTmicrostructureliquidityexecutionpredictionorder-bookhttps://marketmaker.cc/id/blog/koopman-operator-market-dynamics/https://marketmaker.cc/id/blog/koopman-operator-market-dynamics/Dynamic Mode Decomposition fits a linear operator to nonlinear market dynamics. The only question that matters: do the fitted modes persist from one window to the next, and does the rolling spectral radius lead realised volatility? Here is the measurement protocol and the code to run it.Wed, 12 Aug 2026 00:00:00 GMTmathematicsKoopmandynamical-systemsspectralpredictionhttps://marketmaker.cc/id/blog/knowledge-distillation-trading-models/https://marketmaker.cc/id/blog/knowledge-distillation-trading-models/The blog's standing answer to the accuracy-vs-latency tension is a two-stage fast/slow split. Distillation is a different answer: train one small model to mimic the ensemble. The KD loss, temperature, born-again nets, early exits for a variable latency budget, and the distill-to-FPGA pipeline - plus the measurements that would decide whether it beats the two-stage split.Tue, 11 Aug 2026 00:00:00 GMTmodel-compressiondistillationlatencyHFTdeploymenthttps://marketmaker.cc/id/blog/instrumental-variables-quantitative-finance/https://marketmaker.cc/id/blog/instrumental-variables-quantitative-finance/Selection bias in the search is not the only way a regression fools you. When the regressor is correlated with the error, more data makes the estimate more confidently wrong. Instrumental variables applied to the one endogeneity problem this blog has already left open: permanent impact from net taker flow.Mon, 10 Aug 2026 00:00:00 GMTcausal-inferenceinstrumental-variables2SLSeconometricsendogeneityhttps://marketmaker.cc/id/blog/hawkes-process-order-arrival/https://marketmaker.cc/id/blog/hawkes-process-order-arrival/Menyesuaikan proses titik yang menarik dengan rekaman perdagangan kripto nyata: dari mana tiga angka (mu, alfa, beta) berasal, cara memperkirakan rasio percabangan n, apakah kernel eksponensial bertahan dalam uji kesesuaian, dan berapa banyak n bergerak saat Anda mengubah jendela estimasi.Sun, 09 Aug 2026 00:00:00 GMTmicrostructureHawkes-processpoint-processorder-arrivalcryptoHFThttps://marketmaker.cc/id/blog/hamiltonian-neural-networks-finance/https://marketmaker.cc/id/blog/hamiltonian-neural-networks-finance/Jaringan Neural Hamiltonian terbukti stabil - tetapi stabilitas tidak ada gunanya jika kuantitas yang dilestarikan tidak ada. Mempelajari skalar H melalui autograd, integrasi simplektis, dan uji falsifikasi yang memutuskan apakah pasangan finansial (q, p) bersifat kanonik.Sat, 08 Aug 2026 00:00:00 GMTdeep-learningHamiltonianphysics-informeddynamicsconservationhttps://marketmaker.cc/id/blog/granger-causality-lead-lag-crypto/https://marketmaker.cc/id/blog/granger-causality-lead-lag-crypto/Kausalitas Granger pada harga kripto dilakukan dengan dua cara - perbedaan pengembalian dan level Toda-Yamamoto - dengan implementasi Wald yang benar, matriks kausalitas terkoreksi-N yang efektif, dan uji stabilitas bergulir untuk mengetahui apakah lag dapat diperdagangkan atau tidak.Fri, 07 Aug 2026 00:00:00 GMTcausal-inferenceGranger-causalitylead-lagcryptoVARhttps://marketmaker.cc/id/blog/gaussian-processes-trading/https://marketmaker.cc/id/blog/gaussian-processes-trading/Desain kernel untuk deret waktu keuangan - Kekasaran matern, komposisi periodik lokal, campuran spektral - dan kemungkinan marjinal sebagai pengatur yang tidak memerlukan set validasi. Ditambah daftar jujur ​​tentang apa yang masih harus diukur sebelum dapat diperdagangkan.Thu, 06 Aug 2026 00:00:00 GMTbayesiangaussian-processkerneluncertaintynon-parametrichttps://marketmaker.cc/id/blog/fourier-neural-operator-finance/https://marketmaker.cc/id/blog/fourier-neural-operator-finance/Pembelajaran operator memetakan seluruh ruang fungsi, bukan titik - bagaimana Fourier Neural Operator membuat parameter operator solusi PDE dalam ruang frekuensi, apa yang dibeli untuk harga opsi, dan klaim mana yang masih perlu diukur pada perangkat keras sebenarnya.Wed, 05 Aug 2026 00:00:00 GMTdeep-learningFNOPDEoperator-learningoptionshttps://marketmaker.cc/id/blog/flash-attention-trading-models/https://marketmaker.cc/id/blog/flash-attention-trading-models/Flash Attention menjadikan konteks perdagangan 23.400 langkah bebas komputasi - ubin, softmax online, dan batasan IO sebesar N^2 d^2 / M. Apakah konteks yang lebih panjang tersebut menjadikan model lebih baik adalah pertanyaan terpisah, dan harus diukur, bukan ditegaskan.Tue, 04 Aug 2026 00:00:00 GMTdeep-learningattentionFlash-AttentionGPUoptimizationhttps://marketmaker.cc/id/blog/ensemble-methods-trading/https://marketmaker.cc/id/blog/ensemble-methods-trading/Tiga hal tentang ansambel perdagangan yang belum dibahas di blog ini: mengapa istilah kovarians - bukan jumlah model - mendominasi kesalahan ansambel, bagaimana penumpukan berfungsi sebagai lapisan kombinasi sinyal dengan fitur meta yang tidak dapat dilipat, dan apakah jaring omset di banyak sinyal benar-benar mengurangi biaya eksekusi.Mon, 03 Aug 2026 00:00:00 GMTmachine-learningensemblebaggingstackingalphahttps://marketmaker.cc/id/blog/double-ml-causal-trading/https://marketmaker.cc/id/blog/double-ml-causal-trading/Setiap model di blog ini menjawab 'apa yang memprediksi apa?'. Double ML menjawab 'apa yang menyebabkan apa?' - dengan standar error yang bisa Anda pertahankan. Model partially linear, orthogonalitas Neyman, cross-fitting yang dipurges pada data order book, dan perhitungan jujur mengapa interval kepercayaan DML yang valid bertahan untuk tepat satu pertanyaan yang telah ditentukan sebelumnya.Sun, 02 Aug 2026 00:00:00 GMTcausal-inferencedouble-MLtreatment-effecteconometricsquanthttps://marketmaker.cc/id/blog/causal-forests-treatment-effects-trading/https://marketmaker.cc/id/blog/causal-forests-treatment-effects-trading/Setiap backtest di blog ini memperkirakan rata-rata bersyarat. Causal forests memperkirakan efek perawatan bersyarat sebagai gantinya - tau(x) bukan mu(x) - dengan honest splitting, representasi bobot kernel adaptif, dan tes kalibrasi yang memberi tahu Anda apakah heterogenitas yang Anda temukan nyata.Sat, 01 Aug 2026 00:00:00 GMTcausal-inferencecausal-forestheterogeneous-effectseconometricstreatmenthttps://marketmaker.cc/id/blog/bayesian-neural-networks-trading/https://marketmaker.cc/id/blog/bayesian-neural-networks-trading/Setiap aturan sizing di blog ini memperlakukan ketidakpastian sebagai satu angka. MC Dropout dan deep ensembles membaginya menjadi ketidaktahuan model dan noise pasar - dan keduanya layak mendapat ukuran posisi yang berbeda.Fri, 31 Jul 2026 00:00:00 GMTbayesianuncertaintyneural-networkriskposition-sizinghttps://marketmaker.cc/id/blog/automl-systematic-trading/https://marketmaker.cc/id/blog/automl-systematic-trading/Automated feature generation (tsfresh, Featuretools), budget-aware model search (FLAML's cost-frugal optimizer), and the WorldQuant formulaic alpha factory - the parts of the research pipeline this blog's search-and-overfit arc never covered, and what the arc's own results say about them.Thu, 30 Jul 2026 00:00:00 GMTAutoMLNASfeature-engineeringautomationquanthttps://marketmaker.cc/id/blog/liquidation-cascades-trading-signal/https://marketmaker.cc/id/blog/liquidation-cascades-trading-signal/Setiap posisi berleverage mengumumkan harga saat posisi itu harus dijual. Cara membangun grafik kedalaman likuidasi on-chain dan heatmap likuidasi CEX, memodelkan dinamika kaskade sebagai bilangan reproduksi, serta memperdagangkan arus terpaksa sebagai sinyal, bukan hanya menganggapnya sebagai risiko.Wed, 29 Jul 2026 00:00:00 GMTlikuidasikaskadesinyal tradingDeFiperpetualorder bookmikrostruktur pasarkriptohttps://marketmaker.cc/id/blog/uniswap-v3-lp-strategies-hedging/https://marketmaker.cc/id/blog/uniswap-v3-lp-strategies-hedging/An active v3 LP is running a market-making book with gas costs and no cancel button. Size ranges from a GARCH vol forecast, frame rebalancing as fee-gain vs realized-cost, derive the position delta from tick math and hedge it with perps, and understand JIT liquidity and the honest pitfalls of backtesting LP from on-chain data.Tue, 28 Jul 2026 00:00:00 GMTuniswapdefimarket makingdelta hedginglvrjit liquidityquantethereumhttps://marketmaker.cc/id/blog/mev-supply-chain-pbs-mevboost/https://marketmaker.cc/id/blog/mev-supply-chain-pbs-mevboost/MEV bukan lagi sekadar bot tunggal, melainkan telah menjadi jalur perakitan: searcher, builder, relay, proposer. Penjelasan teknis mengenai pemisahan proposer-builder (PBS), lelang blok MEV-Boost dengan tawaran tertutup, mengapa margin searcher hilang dalam lelang, lelang order-flow sebagai parit baru, dan bagaimana model Jito di Solana berbeda.Mon, 27 Jul 2026 00:00:00 GMTmevpbsmev-boostflashbotsjitolelang order-flowdefiethereumhttps://marketmaker.cc/id/blog/onchain-arbitrage-atomic-flash-loans/https://marketmaker.cc/id/blog/onchain-arbitrage-atomic-flash-loans/Menemukan siklus arbitrase on-chain adalah bagian yang mudah. Bagian yang sulit adalah memenangkan lelang prioritas penawaran tertutup melawan setiap pencari lain yang menemukan siklus yang sama. Penjelasan teknis tentang atomicity, flash loan, backrunning, dan di mana keuntungan arbitrase benar-benar bertahan - termasuk CEX-DEX non-atomik.Sun, 26 Jul 2026 00:00:00 GMTarbitrageflash loansmevdefibackrunningcex-dexethereumatomicityhttps://marketmaker.cc/id/blog/impermanent-loss-lvr-lp-profitability/https://marketmaker.cc/id/blog/impermanent-loss-lvr-lp-profitability/Impermanent loss, leverage likuiditas terkonsentrasi, dan loss-versus-rebalancing diturunkan dari prinsip-prinsip pertama. Bentuk tertutup, hasil σ²/8, markout per-blok, dan kondisi biaya-versus-volatilitas yang tepat di mana LP benar-benar mengalahkan HODL.Sat, 25 Jul 2026 00:00:00 GMTdefiammimpermanent losslvruniswapmarket makingvolatilityquanthttps://marketmaker.cc/id/blog/child-order-execution-tactics/https://marketmaker.cc/id/blog/child-order-execution-tactics/Almgren-Chriss dan VWAP hanya menetapkan budget slice - eksekusi dimenangkan di lapisan taktik. Timer eskalasi, matematika break-even maker-taker, semantik queue amend vs cancel-replace di Binance/OKX/CME, anti-signaling iceberg, dan state machine Python per-slice.Fri, 24 Jul 2026 00:00:00 GMTeksekusichild ordertaktik orderposisi antreanmaker-takericeberg ordermikrostrukturpythonhttps://marketmaker.cc/id/blog/slippage-cost-models-backtest/https://marketmaker.cc/id/blog/slippage-cost-models-backtest/Ganti slippage konstan-bps dengan kurva biaya yang bergantung pada ukuran order, volatilitas, dan likuiditas: hukum akar-kuadrat, fitting kurva dari fill TCA atau data publik, pengali stres regime, dan mengapa leaderboard strategi Anda akan tersusun ulang.Thu, 23 Jul 2026 00:00:00 GMTslippagebiaya transaksimarket impactbacktesteksekusitcahukum akar-kuadratmodel biayahttps://marketmaker.cc/id/blog/smart-order-routing-crypto/https://marketmaker.cc/id/blog/smart-order-routing-crypto/Mengapa SOR crypto lebih sulit daripada routing di ekuitas: tidak ada consolidated tape, phantom liquidity, modal yang harus di-prefund. Optimasi routing konveks lengkap dengan matematika dan Python, taktik maker-aware, dan tabel liga markout per venue dari TCA milikmu sendiri.Wed, 22 Jul 2026 00:00:00 GMTeksekusismart order routingmarket microstructurefragmentasi likuiditascryptoquantpythontcahttps://marketmaker.cc/id/blog/maker-taker-fees-rebates-execution/https://marketmaker.cc/id/blog/maker-taker-fees-rebates-execution/Maker vs taker adalah trade adverse selection yang mengenakan kedok jadwal biaya. Matematika titik impas via Glosten-Milgrom, tingkatan biaya kripto nyata dan diskon token, sejarah rebate mining, nilai rebate yang bergantung pada antrean, dan pemodelan biaya bertingkat dalam backtest.Tue, 21 Jul 2026 00:00:00 GMTmaker takerbiaya tradingrebateadverse selectionmarket makingeksekusimarket microstructurebacktesthttps://marketmaker.cc/id/blog/implementation-shortfall-tca-execution/https://marketmaker.cc/id/blog/implementation-shortfall-tca-execution/Implementation shortfall dari Perold diubah menjadi pipeline TCA yang bekerja untuk bot crypto: dekomposisi arrival-price, kurva markout pada t+1s/10s/60s, ~200 baris Python di atas fill Anda sendiri, dan memasukkan hasilnya kembali ke model biaya backtest Anda.Mon, 20 Jul 2026 00:00:00 GMTexecutiontcaimplementation-shortfallmarkoutstransaction-costsalgotradingbacktestmarket-microstructurehttps://marketmaker.cc/id/blog/onchain-liquidations-aave-compound/https://marketmaker.cc/id/blog/onchain-liquidations-aave-compound/Bagaimana likuidasi Aave v3 dan Compound sebenarnya bekerja: matematika health factor, close factor, liquidation bonus, pemicu oracle Chainlink, dan apa yang dibutuhkan untuk menjalankan bot likuidasi flash-loan pada 2026.Sun, 19 Jul 2026 00:00:00 GMTdefilikuidasiaavecompoundmevchainlinkflash loancryptohttps://marketmaker.cc/id/blog/uniswap-v3-concentrated-liquidity-quants/https://marketmaker.cc/id/blog/uniswap-v3-concentrated-liquidity-quants/Bagaimana Uniswap v3 sebenarnya bekerja di balik layar: virtual reserves, liquidity L, sqrtPriceX96, tick math, feeGrowthInside, dan mengapa posisi LP adalah payoff short-volatility. Dilengkapi formula eksak dan contoh numerik yang dikerjakan langkah demi langkah.Sat, 18 Jul 2026 00:00:00 GMTuniswapdefiammlikuiditas terkonsentrasipembuatan pasarlvrquantethereumhttps://marketmaker.cc/id/blog/mev-sandwich-frontrunning-mempool/https://marketmaker.cc/id/blog/mev-sandwich-frontrunning-mempool/Setiap swap yang kamu kirim ke RPC publik adalah limit order yang terlihat dan tidak wajib dihormati siapa pun. Bedah teknis cara kerja ekstraksi MEV: pengurutan mempool, matematika sandwich di Uniswap v2, PGA, Flashbots, dan pertahanan trader.Fri, 17 Jul 2026 00:00:00 GMTmevsandwich attackfrontrunningmempoolflashbotsdefiethereumhttps://marketmaker.cc/id/blog/fill-simulation-partial-fills-backtest/https://marketmaker.cc/id/blog/fill-simulation-partial-fills-backtest/Lima anak tangga fidelitas simulasi fill - dari fill harga close hingga model posisi antrean probabilistik. Partial fill sebagai state machine, batas probabilitas fill limit order sebagai bracket PnL, dan loop kalibrasi terhadap fill live.Thu, 16 Jul 2026 00:00:00 GMTsimulasi fillbacktestlimit orderposisi antreanpartial fillmarket makingeksekusimarket microstructurehttps://marketmaker.cc/id/blog/twap-vwap-pov-execution-algorithms/https://marketmaker.cc/id/blog/twap-vwap-pov-execution-algorithms/TWAP, VWAP dan POV adalah taruhan pada prakiraan volume. Kami membedah asumsi tersembunyi setiap scheduler, membangun kurva volume intraday crypto, dan mengadu ketiganya secara head-to-head pada data L2 yang diputar ulang.Wed, 15 Jul 2026 00:00:00 GMTexecutiontwapvwappovimplementation-shortfallkurva-volumealgotradingbacktesthttps://marketmaker.cc/id/blog/almgren-chriss-optimal-execution/https://marketmaker.cc/id/blog/almgren-chriss-optimal-execution/Derivasi lengkap model eksekusi optimal Almgren-Chriss: dampak linier, trajektori sinh/cosh, efficient frontier, dan kode Python siap pakai untuk mengkalibrasi eta, gamma, dan sigma dari data order book L2 dan trade Binance.Tue, 14 Jul 2026 00:00:00 GMTeksekusialmgren-chrissmarket impacteksekusi optimaltwapquantpythoncryptohttps://marketmaker.cc/id/blog/volatility-targeting-garch-strategy/https://marketmaker.cc/id/blog/volatility-targeting-garch-strategy/Prakiraan volatilitas GARCH baru bernilai jika ia memperbaiki keputusan trading. Kami membangun strategi kripto dengan volatility targeting, mengevaluasi kualitas prakiraan secara jujur, dan membandingkan GARCH dengan baseline realized-vol serta EWMA dalam backtest walk-forward.Mon, 13 Jul 2026 00:00:00 GMTvolatilityGARCHvolatility-targetingbacktestingriskcryptoalgorithmic-tradinghttps://marketmaker.cc/id/blog/dcc-garch-dynamic-correlation-crypto/https://marketmaker.cc/id/blog/dcc-garch-dynamic-correlation-crypto/Korelasi kripto tidaklah konstan - nilainya melonjak ke arah 1 di setiap drawdown. DCC-GARCH memodelkan matriks korelasi yang berubah seiring waktu, memberi Anda hedge ratio dinamis untuk pairs trading serta risiko portofolio yang jujur dan berubah seiring waktu.Sun, 12 Jul 2026 00:00:00 GMTvolatilityGARCHDCCcorrelationportfoliopairs-tradingcryptohttps://marketmaker.cc/id/blog/asymmetric-garch-crypto-leverage/https://marketmaker.cc/id/blog/asymmetric-garch-crypto-leverage/GARCH(1,1) biasa memperlakukan berita baik dan buruk secara sama serta mengasumsikan guncangan Gaussian. EGARCH, GJR-GARCH, dan inovasi Student-t/skew-t memperbaiki keduanya sekaligus memberi Anda VaR dan Expected Shortfall yang jujur untuk kripto.Sat, 11 Jul 2026 00:00:00 GMTvolatilityGARCHEGARCHriskVaRcryptoalgorithmic-tradinghttps://marketmaker.cc/id/blog/garch-volatility-forecasting-crypto/https://marketmaker.cc/id/blog/garch-volatility-forecasting-crypto/Bagaimana model GARCH(1,1) menangkap volatility clustering pada crypto, cara melakukan fitting dengan maximum likelihood menggunakan library arch, dan cara mengubah prakiraan conditional variance menjadi position sizing dan stop dinamis.Fri, 10 Jul 2026 00:00:00 GMTvolatilityGARCHriskforecastingcryptoalgorithmic-tradinghttps://marketmaker.cc/id/blog/honest-negative-no-robust-edge/https://marketmaker.cc/id/blog/honest-negative-no-robust-edge/Puncak dari rangkaian pencarian-dan-overfit, dan berakhir dengan hasil negatif - hasil yang benar. Pencarian dual-timeframe satu simbol pada ETHUSDT menemukan konfigurasi bernilai +16.35% out-of-sample dan +2.62% pada holdout yang belum tersentuh; Deflated Sharpe Ratio, dengan memperhitungkan ~37,000 trial, mendeflasinya menjadi 0.00. Lintasan lintas-instrumen atas lima koin utama (ETH/BTC/SOL/BNB/XRP, masing-masing ~1.18 juta bar 1 menit), diseleksi berdasarkan median out-of-sample, membunuhnya untuk selamanya: DSR dual 0.24 / PBO 0.264, DSR triple 0.14 / PBO 0.327 - keduanya gagal gate. Sang juara profitabel pada 1 dari 5 simbol dan negatif pada sisanya. Inilah tujuan aparatus anti-overfit: mencegah Anda mengirimkan yang terbaik dari noise sebagai alpha.Thu, 09 Jul 2026 00:00:00 GMTalgotradingbacktestoverfittingdeflated-sharpepbohasil-negatifvalidasihttps://marketmaker.cc/id/blog/proving-no-lookahead-multi-timeframe/https://marketmaker.cc/id/blog/proving-no-lookahead-multi-timeframe/Backtest multi-timeframe membocorkan masa depan lewat bar timeframe lebih tinggi yang sedang terbentuk, yang close akhirnya belum ada. Anda tidak bisa meraih keyakinan hanya lewat code review - Anda harus mengujinya. Kami mereproduksi persis aturan closed-bar milik bot live, lalu membuktikan tidak ada kebocoran dengan probe shifted-future: mengganggu setiap bar masa depan dan memastikan setiap sinyal serta trade di masa lalu tidak berubah sama sekali (bitwise unchanged). 25/25 pemeriksaan parity lolos, dan probe ini punya taring.Wed, 08 Jul 2026 00:00:00 GMTalgotradingbacktestbias look-aheadmulti-timeframekebocoran datavalidasihttps://marketmaker.cc/id/blog/when-gpu-pays-off-sweep-roofline/https://marketmaker.cc/id/blog/when-gpu-pays-off-sweep-roofline/Keunggulan GPU atas CPU tumbuh seiring ukuran batch - 54.5x pada satu kombo per panggilan hingga 359.6x pada 61 di precompute indikator multi-timeframe kami - karena sweep kecil tidak bisa mengamortisasi overhead peluncuran kernel dan transfer. Kami mengurai headline 167x menjadi kemenangan algoritmik 27x yang juga membantu CPU kali kemenangan hardware 6.2x, menunjukkan bahwa keunggulan sejati GPU-vs-CPU-terbaik hanya 3.2x untuk single-timeframe dan 6.2x untuk multi, dan memberikan panduan keputusan seberapa lebar sweep yang dibutuhkan sebelum GPU layak dibeli.Tue, 07 Jul 2026 00:00:00 GMTalgotradingbacktestperformagpurooflineoptimasi