VALYNOX
VALYNOXSTRATEGY RESEARCH TECHNOLOGY
AlphaPrimeAlphaPrime
ACADEMY / Platforms and export

Yuanta vs Capital Futures MultiCharts: What Differs

The Yuanta and Capital Futures MultiCharts programs and backtest engines are the same. The differences are the broker-version license and connection, quote source and history length, order API, and fee terms. Data length and bar cut change the backtest numbers of the same strategy, while the order path and commissions decide the gap between live results and the backtest.

By · Published · updated · Editorial policy

Contents
  1. How do Yuanta and Capital Futures MultiCharts differ?
  2. How do the two brokers' versions connect, and what does each provide?
  3. How long is Yuanta MultiCharts history, and do quotes change backtest results?
  4. How do automated orders and order paths differ?
  5. How much are Capital Futures MultiCharts commissions, and what should you confirm before opening an account?
  6. Frequently asked questions
  7. References

How do Yuanta and Capital Futures MultiCharts differ?

Both brokers' MultiCharts share the same program and backtest engine. The differences are the license and connection of the broker version, the quote source and how far back history reaches, which broker API carries automated orders, and commission and margin terms. Quotes and data length change the backtest numbers of the same strategy directly. Commissions and the order path decide how far live fill prices drift from the backtest's assumptions.

So the real question when comparing the two is this: how much of my backtest number comes from the data, and how much from the rules? A backtest report is the combined result of “this data, this rule and this selected candidate.” Change any one of them and the numbers change. The rest of this article follows that order: connection, data, orders, fees.

How do the two brokers' versions connect, and what does each provide?

A broker version of MultiCharts is MultiCharts licensed through a broker's quote and order connection. Quotes, history and orders all go through components that broker provides. Yuanta and Capital Futures each have their own connection component and account conditions, so follow each broker's notices for feature details. The connection steps are broadly the same for both:

  1. Open a futures account and apply to your broker representative for MultiCharts access. Application conditions (for example, whether electronic trading or the API must be enabled) follow the broker's notices.
  2. Install MultiCharts and the broker's connection component, then log in to the component with your futures account number.
  3. Add the broker's quote source in QuoteManager and load the symbol you want to test, then note the date of the earliest bar actually retrieved.
  4. Place a test order with the smallest contract size, and confirm that order submission, fill reports and position sync all work.

Yuanta's step-by-step setup and quote checks are in Yuanta MultiCharts connection and quote checks. KGI Futures' setup is in KGI MultiCharts account, API and test order. Capital Futures' process follows the same outline as those two; only the component and menu names differ.

ItemYuantaCapital Futures
Program and backtest engineSameSame
License sourceBroker version, per broker noticeBroker version, per broker notice
Connection componentYuanta quote and order APICapital Futures quote and order API
History length and sessionsConfirm with the brokerConfirm with the broker
Automated ordersRequires broker API and account activationRequires broker API and account activation
Commission and marginConfirm with your broker representativeConfirm with your broker representative
Yuanta and Capital Futures use the same MultiCharts program and backtest engine; quote source, order path and fees each need to be confirmed with the broker, so backtest gaps come from data, not software Same engine, different sources Yuanta version vs Capital Futures version Engine Same on both Data Per broker Orders Per broker Fees Per broker The gap comes from data, not software. So check data length first.
Same program; what differs is who supplies the connection, data and order path. Items marked “Same” need no comparison. Confirm the items marked “per broker notice” with each broker one by one.

How long is Yuanta MultiCharts history, and do quotes change backtest results?

How far history reaches and which timestamp cuts the bars decide whether the same rule fires the same signal on the two brokers' data, so yes, the two brokers' quotes do change backtest results. History length is the span from the earliest bar the quote source provides to today. A timestamp is the time attached to each trade, and the bar groups the trade into a minute accordingly. To see how far back each broker goes, look at the earliest bar directly after loading a symbol, and ask the broker whether the night session is included and how gaps are filled. The numbers change as brokers adjust, so this article gives no fixed value.

The cut matters in a subtler way than length. The TAIEX futures (TX) day session runs 08:45 to 13:45 (the exchange may adjust it, so follow the TAIFEX contract page; you can check each session in the futures market hours tool). If the same stream of trades carries timestamps a few seconds apart on the two brokers, or one broker folds the first trade at 08:45 into the previous bar, the open, high, low and close of a 5-minute bar differ, and a moving-average crossover may trigger on one and not on the other. Dozens of such differences add up, and the number of trades and net profit drift apart.

Illustration: the same trades fall after the 10:35 bar boundary in Quote A and before it in Quote B, so Quote A shows a crossover signal and Quote B does not Same trades, different signal A batch of trades spans 10:35 Schematic Quote A Signal Quote B No signal 10:35 Trades Shift the bar cut, flip the signal. So align sessions before comparing.
Same price, different bar cut, different signal. Moving the bar boundary shifts only one batch of trades, and the 10:35 crossover appears in only one of the rows (illustration).

We ran a demonstration with AlphaPrime (Valynox) to show that the same result changes when you swap the data. The run uses NQ daily bars, 2019-01-02 to 2025-06-30, long and short, 1 contract, with a single linear regression rule for entries and exits, and no stop loss, target or filter. It is a stand-in rule built from standard indicators, not the topic of this article. The row in the table is the one kept, ranked by return relative to drawdown, after searching 1,024 candidates. A candidate picked after the fact looks better than average by construction, and Bailey et al. (2014, 2017) describe exactly this overestimate from selection, so these numbers are not what this rule earned and not a result you should expect.

Source: AlphaPrime product engine (CPU), run 2026-10-31. NQ daily, 2019-01-02 to 2025-06-30, long and short, 1 contract, commission $20 per trade, slippage $10.

NumberResult
Net profit (closed trades)$150,865
Max drawdown$111,615
Trades127
Win rate53.54%
Profit factor1.355
Average trade$1,187.91
Time in market86.09%
Read four things before net profit: only 127 trades, NQ daily bars from 2019 to 2025, costs of US$20 commission plus US$10 slippage, and one candidate picked from more than a thousand Numbers from this run Check these before net profit Data This broker's bars Rule One exported rule Candidate Best of many tried = Report Valid for this mix only Swap one part, the numbers move. So: rerun on your broker's data.
Check the number of trades, the period and the cost settings. These figures are tied to one data set and one selected candidate, not to a property of the rule itself. NQ · daily · 2019-01-02 to 2025-06-30 · CPU engine backtest

127 trades with 86.09% time in market means the rule is almost always in a position, so a difference in the open or close of a single daily bar can move the entry and exit dates. If you export the same rule to PowerLanguage and run it on Yuanta or Capital Futures data, following the steps for importing into MultiCharts and re-running the backtest, the numbers will differ from the table. The difference comes from the data, not the rule. The platforms AlphaPrime exports to are listed on strategy export platforms.

How do automated orders and order paths differ?

Automated order differences do not affect the backtest, but they decide how far live fills sit from the backtest's assumptions. Both brokers send MultiCharts signals to the market through their own order API. Compare three things:

  1. Connection stability: how many times a day the API disconnects, whether it reconnects automatically, and whether positions stay in sync with MultiCharts.
  2. Supported order types: whether market, limit and stop orders can all be sent, and whether a stop order is triggered by the broker's server or by your local computer. The outcome is completely different if the computer crashes.
  3. Report speed: the delay before fill reports reach MultiCharts determines whether the next signal is based on the correct position.

Measure each of these with a test order at the smallest contract size, and record the gap between signal price and fill price. That gap is the slippage to enter in the backtest, not a number you guess.

How much are Capital Futures MultiCharts commissions, and what should you confirm before opening an account?

Commissions, margin and license terms are adjusted by the broker from time to time, so this article lists no fixed numbers. Confirm each item with your broker representative before opening an account, then enter the confirmed costs in the backtest. How costs are composed and the general settings are covered in MultiCharts Taiwan fees and first strategy tutorial. Confirm in this order:

  1. License and account conditions: MultiCharts access, any trading volume or account balance requirement, and the API activation process.
  2. History length and sessions: the earliest bar date, whether the night session is included, and how history is filled in.
  3. Quote timestamps and bar cut: where timestamps come from, and which bar the first trade at the open belongs to.
  4. Order API and order types: where stop orders are triggered, reconnection after a disconnect, and report delay.
  5. Commission and margin: the commission per contract, exchange fee and futures transaction tax added up as the round-trip cost, plus day-trade and overnight margin. Enter these in the backtest's commission and slippage fields.
Confirm five items with the broker, in order: license terms, history and sessions, timestamps and bars, order API and order types, commission and margin. Item five feeds the backtest's commission and slippage settings Confirm five things first Before picking a version 1 2 3 4 5 License terms History, sessions Timestamps, bars Order API, types Fees and margin Ask the broker Backtest Fees, slippage Item five sets backtest cost. So: enter costs before searching.
The order of checks before picking a version: confirm all five items with the broker, and enter the fifth directly as the backtest's commission and slippage.

The number from step five is used in every evaluation, not subtracted after the search ends. Once you enter the round-trip cost your broker representative quoted and the slippage measured by test orders into the backtest settings in AlphaPrime, every candidate strategy carries that cost and is scored with fills on the next bar, so the ranking is already net of costs.

Next step: load TAIEX futures (TX) at the broker where you plan to open an account, note the date of the earliest bar and which bar the first trade at the open belongs to, then place a test order at the smallest contract size and write down the gap between signal price and fill price. These two numbers are the data boundary and the slippage for your first backtest.

Frequently asked questions

How is a broker version of MultiCharts different from MultiCharts bought from the vendor?

A broker version gets its usage rights through that broker's connection component, and quotes, history and orders all go through that broker. A version bought from the vendor lets you choose your own data source and order connection. The program and backtest engine are the same. What differs is who supplies data and orders. For details, follow the notices from MultiCharts and the broker.

The same strategy gives different backtest results on Yuanta and Capital Futures. Which one is correct?

Both are correct, each describing its own data. Differences usually come from history length, timestamps and bar cut, not from the program calculating wrongly. Align the backtest start and end dates and the trading session first, then compare. What remains is the difference between the two brokers' data. Whichever broker's data you actually trade on, use that broker's result.

Do I need to re-run the backtest after switching brokers?

Yes. When the data source changes, bars and history length can change, and entry and exit dates move with them. When you re-run, enter the new broker's commission and the slippage you measured, then compare the number of trades, profit factor and maximum drawdown against the old results. A large gap means the strategy is sensitive to how the data is cut.

How do I enter commissions in a backtest?

Add up the broker commission, exchange fee and futures transaction tax into one round-trip cost per contract and enter it in the backtest's commission field. For slippage, use the gap between signal price and fill price from your test orders, entered in points or as an amount. Put both in from the first run of the search and backtest, so the candidate ranking reflects results net of costs.

References

  • Bailey, D. H., Borwein, J., López de Prado, M., & Zhu, Q. J. (2014). Pseudo-Mathematics and Financial Charlatanism: The Effects of Backtest Overfitting on Out-of-Sample Performance. Notices of the American Mathematical Society, 61(5), 458–471.
  • Bailey, D. H., Borwein, J., López de Prado, M., & Zhu, Q. J. (2017). The Probability of Backtest Overfitting. Journal of Computational Finance, 20(4), 39–69.
  • Taiwan Futures Exchange. Mini-TAIEX Futures. https://www.taifex.com.tw/enl/eng2/mTX