{
    "release": "v2.0",
    "created": "2026-08-23",
    "title": "The 2011 MtGox series",
    "summary": "MtGox BTC/USD top of book and weighted depth scores at four to five second cadence through the autumn 2011 crash, in two runs across a month.",
    "license": "CC-BY-4.0",
    "citation": "bitcoinpricedata.com (2026). The 2011 MtGox series, v2.0. https://bitcoinpricedata.com/data/v2.0/",
    "doi": null,
    "era": "hist2011",
    "source": {
        "artifact": "historical-2011.sqlite",
        "artifact_bytes": 10887168,
        "artifact_sha256": "201c0207f86875599959f98affbc256b0c08f02a499d0c3fbeaf447edc470edc",
        "dataset": "bitcoinpricedata.com historical series, MtGox BTC/USD top of book, autumn 2011",
        "release": "v2.0",
        "era": "hist2011",
        "import_date": "2026-08-23",
        "generated_by": "bayesdata scripts/import-2011.php",
        "source_file": "/home/carlos/code/arbitrage-bots/arbitrage/_not_used/isaac-old-data.sql (14135038 bytes), mysqldump of database `connected`, table `isaacbotdata`, taken on the collecting machine 2011-10-20, 14135038 bytes of SQL read by this import",
        "source_sha256": "f933364e1d3f8d309f846bd92902afc52227c08017076069109ad78279676741",
        "source_archive": "The same bytes survive at /mnt/storage/archive/machines/quen/knowhow/backup_bitcash/arbitrage/_not_used/, which is where the committed repository copy came from. Recovered and identified 2026-08-23.",
        "venue": "MtGox, mtgox.com API, BTC/USD. The exchange failed in February 2014 and this book cannot be collected again at any price.",
        "quote_currency": "USD",
        "collector": "bot.isaac.php with core.isaacbot.php, a trading bot rather than a collector. Source preserved at ~/code/arbitrage-bots/arbitrage/_not_used/. It logged one row every third pass of its main loop, so the cadence is the loop's and not a schedule.",
        "cadence": "4 s on 74.0%, 5 s on 24.5%, 6 s on 0.7%, 7 s on 0.3% of 177583 in-run steps. The interval follows how long MtGox took to answer, so treat the series as irregular and key on the timestamp, never on row position.",
        "time_column": "dateline, unix seconds UTC, PRIMARY KEY, one row per second at most",
        "price_unit": "bid, ask and mean are US dollars per BTC, as MtGox quoted them. No conversion of any kind was applied: the source columns are float(6,4) and the artifact carries the same four decimals the dump printed.",
        "columns": "dateline, bid, ask (MtGox best bid and best ask), mean (their midpoint as the bot computed and stored it, which is the only price this series has, there is no trade price anywhere in it), spread, bidscore, askscore, diff.",
        "depth_score_definition": "THE 2011 DEPTH SCORES ARE NOT A RATIO, AND NOT THE 2014 OR MODERN ONE. bidscore and askscore each sum the resting volume within $0.50 of the best price on their side of the book, weighting every level by 1/(1 + |price - best|)^2 and flooring the sum to an integer. Read off getdisplayscore() in core.isaacbot.php, called with maxdiff = 0.5. At the $2 to $6 of this era a fifty cent band is 8% to 25% of price, which has no counterpart in the 2014 series (top sixty levels a side) or the modern one (a +/-2% band). Three eras, three depth definitions, never one series.",
        "diff_definition": "diff is the natural log of the ratio of the larger score to the smaller, signed so that it is NEGATIVE when the bid side is the heavier one. That is the opposite sign convention from the 2014 and modern book_ratio columns, where positive is bid-heavy, and it is an unbounded log ratio rather than a -1 to +1 band ratio. It is not published as a ratio anywhere and must not be plotted as one. It is stored as the bot recorded it, at three decimals. Recomputing it from the two stored scores lands within a few thousandths but not exactly, so the stored column is the record.",
        "excluded_columns": "The source table also holds lastprice, closeposition and target. Those are the bot's own trading state, zero on every row where it held no position, and they are not market data. They were dropped at import and are not part of this artifact. They remain in the arbitrage-bots repository.",
        "precision_note": "The source columns are MySQL float(6,4) for the four prices and float(5,3) for diff: single precision, rounded to a fixed number of decimals on store. The four decimals in this file are therefore the full precision that was ever recorded, not a truncation applied later, and there is nothing finer to recover. This import parsed the dump text rather than querying a restored copy, so the digits here are the digits mysqldump printed.",
        "zeros_policy": "This collector wrote no row at all when the MtGox depth call failed, so there are no fabricated placeholder rows of the kind the 2014 series carries. bid, ask, mean, bidscore and askscore are non-zero on every row. spread is 0.0000 on 4776 rows and diff is 0.000 on 44 rows, and both are readings rather than failures: a locked or all but locked book, and two depth scores that came out level.",
        "spread_note": "spread is stored as the bot computed it, abs(bid - ask) formatted to four decimals from the raw book prices, not from the rounded bid and ask in this table. On a few hundred rows it therefore reads 0.0000 where ask - bid works out to 0.0001, and on 7 rows the stored ask is below the stored bid. Recompute the spread from bid and ask if the derived value matters more than the recorded one.",
        "runs": "Two runs. 2011-09-21 13:59:26 to 2011-09-30 05:10:12 UTC, 173730 rows. Then 2011-10-19 19:25:48 to 2011-10-20 01:18:19 UTC, 3855 rows. The bot was not running in between, so the hole is an absence of collection and not an absence of market.",
        "hole.1": "2011-09-30 05:10:12 UTC, 1692936 s (19.6 days) with no rows",
        "gaps_in_run": "Stalls of 60 s or more inside a run, 18 of them, each given as the last timestamp before the stall and its width: 2011-09-21 21:54:40 UTC for 67 s, 2011-09-29 23:41:43 UTC for 187 s, 2011-09-30 01:54:49 UTC for 259 s, 2011-10-19 19:43:12 UTC for 97 s, 2011-10-19 20:50:25 UTC for 193 s, 2011-10-19 20:54:10 UTC for 97 s, 2011-10-19 21:06:14 UTC for 97 s, 2011-10-19 21:31:50 UTC for 97 s, 2011-10-19 21:38:19 UTC for 97 s, 2011-10-19 22:37:52 UTC for 98 s, 2011-10-19 23:14:44 UTC for 98 s, 2011-10-20 00:18:33 UTC for 97 s, 2011-10-20 00:26:32 UTC for 194 s, 2011-10-20 00:31:29 UTC for 97 s, 2011-10-20 00:43:11 UTC for 284 s, 2011-10-20 00:49:01 UTC for 194 s, 2011-10-20 00:56:04 UTC for 314 s, 2011-10-20 01:01:43 UTC for 209 s.",
        "trade_log_note": "The bot's own trade log (table isaacbotsignal, 48 trades between 2011-09-21 and 2011-09-26) survived in a separate 2022 forum database dump. It is the operator's private trading record, it is not market data, and it is not published.",
        "license": "CC BY 4.0",
        "documentation": "bayesdata docs/data-inventory.md, \"The 2011 isaac series\", and docs/spec-2011-import.md",
        "rows_total": "177585",
        "tables": "mtgox",
        "chart_price_note": "The chart JSON for this era carries mean as its price field. There is no trade price in this series at all, only the book midpoint, and every published surface says so."
    },
    "tables": [
        {
            "name": "mtgox",
            "venue": "MtGox, BTC/USD (bankrupt since 2014)",
            "venue_id": "mtgox",
            "quote": "USD",
            "rows": 177585,
            "first_dateline": 1316613566,
            "last_dateline": 1319073499,
            "first_utc": "2011-09-21 13:59:26",
            "last_utc": "2011-10-20 01:18:19",
            "columns": [
                "dateline",
                "bid",
                "ask",
                "mean",
                "spread",
                "bidscore",
                "askscore",
                "diff"
            ],
            "note": "MtGox best bid and ask, their midpoint and spread, plus two distance-weighted depth scores. bidscore and askscore sum book volume within $0.50 of the best price on their side, weighted by 1/(1+|price-best|)^2 and floored, and diff is their log ratio. THAT IS A THIRD DEPTH DEFINITION, unrelated to the 2014 top-60 ratio and to the modern +/-2% band: three eras, three definitions, never one series. Prices come from single-precision float(6,4) storage, so they carry four decimals. There is no trade price in this series, only the book midpoint.",
            "files": [
                {
                    "name": "mtgox.csv.gz",
                    "bytes": 910859,
                    "sha256": "c81955217072c4fdc5c0450aad4196ab62abd646a0908ee733de462d82c37d08"
                },
                {
                    "name": "mtgox.parquet",
                    "bytes": 1381517,
                    "sha256": "3d7258fc15df574d8cba035d57adc0d118a7cf7aa9ad0ea80adee82030dcf4f0"
                }
            ]
        }
    ]
}
