Backfill Medved Trader's charts and history from data files on your own disk — in CSV, TAB, JSON or XML — instead of from a data feed.

LOCAL Backfill is a data source like any other, except that it reads plain files from a folder you choose. Point it at a folder, drop in a file per symbol, and MT will merge that data into its own store the next time it backfills that symbol. It's the way to get data into MT that no feed of yours carries: a symbol your provider doesn't cover, history older than your feed goes back, tick data exported from another application, or data you generate yourself.

LOCAL Backfill only supplies backfill. It never streams live quotes, and it never writes to or deletes your files.

Setting it up

Go to Settings → Data Sources → Configure Accounts and select LOCAL Backfill.

1. Set Backfill File FOLDER Path to the folder holding your data files. There's no login — the folder is the entire configuration.
2. Optionally adjust Intraday OHLC Backfill – Number of Days (default 120, maximum 120) and Intraday TICK Backfill – Number of Days (default 5, maximum 120).
3. Assign LOCAL Backfill as the backfill source for the symbols you want it to serve, the same way you would any other source.

How your files are named

One file per symbol, per kind of data. The name tells MT both things:

{type}_{SYMBOL}.{ext}

Prefix	Holds	Example
ohlc_	Intraday bars	ohlc_AAPL.csv
tick_	Individual trades and quote changes	tick_MSFT.json
hist_	Daily end-of-day bars	hist_SPY.xml

The extension picks the format: csv, tab, json or xml. MT tries them in that order and uses the first file it finds, so don't keep two formats for the same symbol and type.

Use the symbol exactly as MT shows it — the string you'd type into a symbol box, including any type character: AAPL, $SPX, @ESZ25, .TSLA_260515C800. There is no symbol column inside the files; if one is present it's ignored.

Symbols containing a character Windows won't allow in a file name are adjusted: a colon becomes an underscore, and any other invalid character becomes an exclamation mark. So RY:TSX is read from ohlc_RY_TSX.csv.

Rules that apply to every format

• One file per symbol, not per day. Append to it or rewrite it as you please. MT reads the whole file each time it backfills and merges the contents in, so repeating rows you've already supplied is harmless.
• Field names decide the layout, not column order. CSV and TAB files need a header row; MT finds each column by its name. Names are case-insensitive and columns MT doesn't recognize are ignored, so extra columns can stay.
• Row order can go either way. Oldest-first or newest-first are both fine — MT detects the direction and flips the data if needed. Don't mix directions within one file.
• Volume is always optional.
• A price of 0 means "I don't have this", not "the price was zero". Leave a field out or set it to 0 and MT fills it in: a missing open becomes the previous bar's close, and a missing high or low becomes that bar's close.
• Files are UTF-8, and CSV/TAB have no quoted-field handling — so no delimiters inside a value, and no thousands separators inside a number.

Timestamps

Any date or time that .NET can parse will work, but ISO 8601 is what you want: 2026-08-13T09:30:00, with milliseconds for ticks (2026-08-13T09:30:00.100). Unix/epoch numbers are not supported.

In CSV and TAB files the timestamp can be one datetime column or a separate date and time pair — MT works out which from the header and joins the two halves itself.

OHLC — intraday bars

ohlc_SYMBOL.csv  .tab  .json  .xml

Field	Required	Notes
datetime	Yes	Or date + time as two columns. UTC.
close	Yes
open	No	Defaults to the previous bar's close
high, low	No	Default to that bar's close
volume	No	vol is also accepted

CSV / TAB

datetime,open,high,low,close,volume
2024-01-02T09:30:00,185.50,186.20,185.10,185.90,123456
2024-01-02T09:31:00,185.90,186.50,185.75,186.30,98765

Or with the date and time in separate columns:

Date,Time,Open,High,Low,Close,Volume
06/10/26,09:30:00,185.50,186.20,185.10,185.90,123456
06/10/26,09:31:00,185.90,186.50,185.75,186.30,98765

JSON

An array of objects. Long or short field names, detected per row:

[
  {"datetime":"2024-01-02T09:30:00","open":185.50,"high":186.20,"low":185.10,"close":185.90,"volume":123456},
  {"datetime":"2024-01-02T09:31:00","open":185.90,"high":186.50,"low":185.75,"close":186.30,"volume":98765}
]

[
  {"t":"2024-01-02T09:30:00","o":185.50,"h":186.20,"l":185.10,"c":185.90,"v":123456},
  {"t":"2024-01-02T09:31:00","o":185.90,"h":186.50,"l":185.75,"c":186.30,"v":98765}
]

XML

<bars>
  <bar><datetime>2024-01-02T09:30:00</datetime><open>185.50</open><high>186.20</high><low>185.10</low><close>185.90</close><volume>123456</volume></bar>
</bars>

TICK — trades and quote changes

tick_SYMBOL.csv  .tab  .json  .xml

Field	Required	Notes
datetime	Yes	Or date + time as two columns. UTC.
price, volume	No	The trade
bid, ask	No	The quote

A record can be a trade, a quote change, or both — omit or zero whatever doesn't apply. A row with price and volume both 0 but a bid or ask present is treated as a quote-only tick: the bid/ask updates and no trade is recorded. That's how you supply bid/ask history.

CSV / TAB

datetime,price,volume,bid,ask
2024-01-02T09:30:00.100,185.50,200,185.49,185.51

Or with the date and time in separate columns:

Date,Time,Price,Bid,Ask,Volume
06/10/26,14:15:36,185.50,185.49,185.51,200
06/10/26,14:15:37,0,185.50,185.52,0

JSON

[
  {"datetime":"2024-01-02T09:30:00.100","price":185.50,"volume":200,"bid":185.49,"ask":185.51},
  {"datetime":"2024-01-02T09:30:00.200","price":0,"volume":0,"bid":185.50,"ask":185.52},
  {"datetime":"2024-01-02T09:30:00.300","price":185.51,"volume":100}
]

Short names work here too — note that price is c:

[
  {"t":"2024-01-02T09:30:00.100","c":185.50,"v":200,"b":185.49,"a":185.51},
  {"t":"2024-01-02T09:30:00.200","c":0,"v":0,"b":185.50,"a":185.52}
]

XML

<ticks>
  <tick><datetime>2024-01-02T09:30:00.100</datetime><price>185.50</price><volume>200</volume><bid>185.49</bid><ask>185.51</ask></tick>
  <tick><datetime>2024-01-02T09:30:00.200</datetime><price>0</price><volume>0</volume><bid>185.50</bid><ask>185.52</ask></tick>
</ticks>

HIST — daily bars

hist_SYMBOL.csv  .tab  .json  .xml

Field	Required	Notes
date	Yes	A plain trading date — no time, no time zone
close	Yes
open	No	Defaults to the previous day's close
high, low	No	Default to that day's close
volume	No

Daily history is the one place with no day limit — supply as much as you have.

CSV / TAB

date,open,high,low,close,volume
2023-12-29,185.00,186.50,184.75,185.20,45678900
2024-01-02,185.20,187.10,184.90,186.80,52345600

JSON

Use date here rather than datetime:

[
  {"date":"2023-12-29","open":185.00,"high":186.50,"low":184.75,"close":185.20,"volume":45678900},
  {"date":"2024-01-02","open":185.20,"high":187.10,"low":184.90,"close":186.80,"volume":52345600}
]

[
  {"t":"2023-12-29","o":185.00,"h":186.50,"l":184.75,"c":185.20,"v":45678900},
  {"t":"2024-01-02","o":185.20,"h":187.10,"l":184.90,"c":186.80,"v":52345600}
]

XML

The root is <history> and rows use <date>, not <datetime>:

<history>
  <bar><date>2023-12-29</date><open>185.00</open><high>186.50</high><low>184.75</low><close>185.20</close><volume>45678900</volume></bar>
</history>

Limits

Data	How far back
Intraday OHLC	Your Intraday OHLC Backfill setting — up to 120 days
Ticks	Your Intraday TICK Backfill setting — default 5 days, up to 120
Daily history	No limit

Rows older than those windows are read but not kept. Medved Trader's own limit on how many days of tick data it stores applies on top of this, so raising the backfill setting alone won't make MT hold more intraday data than it's configured to store.

There's no cap on file size or row count, but a file is read into memory whole — a multi-hundred-megabyte tick file will cost you the memory to match. Symbols are backfilled one at a time, up to 20 at once.

What MT does to your files

Nothing. Files are opened, read, and left exactly as they were — never moved, renamed, rewritten or deleted. Every backfill re-reads the file from disk, so updating a file is enough to make the new data available; there's no cache to clear and no import step to repeat.

If a symbol comes back empty

Symptom	Usually
Nothing loads for a symbol	The file name doesn't match. The log records the exact name MT looked for — compare it against what's in the folder, watching for the symbol's type character and the character substitutions above.
A CSV or TAB file loads nothing	A required column is missing or misspelled in the header row. The log names the column it couldn't find.
Some rows are missing	Rows whose date can't be parsed, and rows dated before 1901, are skipped silently. Check the format of the timestamps in the rows that vanished.
Bars land at the wrong time of day	The timestamps aren't UTC. See Timestamps above.
Bars are there but flat or missing prices	A price of 0 is read as "not supplied" and filled in from the close. Supply real values, or leave the field out deliberately.
Data older than a few days never appears	The backfill day settings, or MT's tick storage limit, are trimming it. See Limits.