Four in 10 NinjaTrader Futures Traders Use AI, but Few Let It Place Trades
Four in 10 NinjaTrader Futures Traders Use AI, but Few Let It Place Trades
Not financial advice. Past performance is not indicative of future results. Trading involves substantial risk of loss. Do your own research before making any investment decisions. See our Editorial Policy for details on how we test and rate AI trading bots and algorithmic platforms.
NinjaTrader's first Futures Trader Pulse Survey puts a number on something we have watched build for two years: 40 percent of the broker's active clients now use AI somewhere in their trading. The survey, run by email in September across 511 clients who had placed a trade in the prior three months, sits squarely in the AI trading bot and algorithmic trading platform category, the same space we benchmark every quarter Finance Magnates. The headline number is not the interesting one. The interesting one is 22 percent, the share of those AI users who let the software actually place the order.
We have benchmarked against Zephyr AI's adaptive engine in our 2026 review cycle, and the pattern NinjaTrader found matches what we see in funded-account tests. Traders adopt AI for research long before they hand it the execution button. That gap, not the 40 percent headline, is where the money and the risk actually live.
What the NinjaTrader AI survey actually measured
The survey is a client poll, not a market study, and that distinction matters for how we read it. NinjaTrader polled 511 clients by email in September, and only clients who had traded in the previous three months were eligible. Everyone who answered chose to answer, so the results describe NinjaTrader's own client base, not the futures market at large [Finance Magnates, 2026]. We treat self-selected broker polls as directional, not definitive. The 40 percent figure is a floor for engagement, not a ceiling for adoption.
The breakdown inside that 40 percent is where the signal lives. Among NinjaTrader clients who use AI, 84 percent apply it to researching or developing strategies, and 37 percent take live trade signals or entry and exit recommendations from it. Only 22 percent of AI users let it execute trades automatically. By Finance Magnates' own calculation, that is roughly 45 of the 511 respondents, or about 9 percent of the whole sample, handing execution to software [Finance Magnates, 2026].
So the real ratio in this data set is four to one. For every client letting AI place trades, roughly four use it only to think. That is the number we would put in front of anyone shopping for an AI trading bot in 2026, and it is a far more honest starting point than the "AI is taking over trading" framing that dominates most coverage.
Why do 78 percent of AI users still place trades by hand?
Because the cost of a bad research call is a wasted afternoon, and the cost of a bad execution call is a margin call. The survey does not ask why, but the adjacent numbers hint at it. Twenty-nine percent of NinjaTrader clients said they were interested in AI but not yet using it, and 31 percent had no plans to [Finance Magnates, 2026]. That is 60 percent of the client base either on the fence or off it entirely.
We have logged the same behavior in our own evaluation framework. Traders will run a strategy through our backtest harness, read the equity curve, and still refuse to connect it to a live account until they have watched it for a full quarter. That is rational. The gap between a backtest and a funded account is where most AI trading claims die, and the 22 percent execution figure is the market pricing that gap in real time.
Compare NinjaTrader's 40 percent AI-use figure with the CMC Markets poll of 8,506 Australian clients, published in September, which found 48.6 percent used AI to support investment decisions while only 29.4 percent trusted AI-generated market insights [Finance Magnates, 2026]. NinjaTrader's adoption number sits 8.6 percentage points below CMC's, but the trust gap is the more portable finding. Roughly a third of AI users in both samples do not trust the output they are already using.
What two broker polls say about AI adoption
| Metric | NinjaTrader Pulse Survey | CMC Markets poll |
|---|---|---|
| Sample size | 511 active clients | 8,506 Australian clients |
| Fieldwork | Email, September | Published September |
| Use AI to support decisions | 40% | 48.6% |
| Trust AI-generated insights | Not reported | 29.4% |
| Let AI execute automatically | 22% of AI users (about 9% of full sample) | Not reported |
| Interested but not using | 29% | Not reported |
| No plans to use | 31% | Not reported |
Source: Finance Magnates, NinjaTrader Futures Trader Pulse Survey and CMC Markets client poll, 2026. Cells marked "Not reported" were not disclosed in the source material.
The gap between research AI and execution AI
Here is the part the headline misses, and it is the part that matters most to anyone evaluating an AI trading bot. "Using AI" in this survey is a wide bucket. It includes a trader pasting price data into a chatbot, a trader running a machine-learning classifier on tick history, and a trader wiring an agent to a live order API. Those are three different products with three different risk profiles, and the survey folds them into one 40 percent figure.
When an AI agent can call external tools, through something like the Model Context Protocol that NinjaTrader's new chief innovation and AI officer has a remit over [Finance Magnates, 2026], the strategy stops being a specification you can re-implement and becomes a non-deterministic process you can only sample. That is the under-discussed risk. A rules-based expert advisor has a fixed decision tree, so we can re-implement it line by line, run it across a multi-year window, and check whether the vendor's published equity curve is reproducible. A natural-language agent that decides which tool to call does not give you that. You cannot walk-forward a prompt.
This is why we separate machine learning from rule-based automation in every review. A gradient-boosted model that outputs a position size is still, at the execution layer, a rule: if score is above a threshold, then trade N contracts. That is testable. An LLM that reasons about the market and then chooses whether to place an order is not, at least not with the tooling most retail traders have in 2026. The 84 percent research share in NinjaTrader's data is mostly the untestable kind. The 22 percent execution share is where the testable kind lives. Trading software vendors such as B2Broker now let AI agents place orders, with brokers setting the limits [Finance Magnates, 2026], and that model pushes even more of the decision into the non-deterministic layer.
What happens when a bot deviates from its spec?
This is the question that separates a serious evaluation from a marketing page, and it is the one the survey cannot answer. In our review process, we read the strategy file where the vendor ships one, and we flag any behavior that is not in the published specification. Common deviations include an undocumented stop-loss override that triggers on a volatility spike, a position-size cap that quietly resets after a losing streak, and a session filter that disables trading during a news window the spec never mentions.
We treat every deviation as a potential source of backtest-to-live divergence. If the spec says the bot trades 24 hours and the code disables entries for 30 minutes around a scheduled data release, the live results will not match the backtest, and the gap will look like bad luck when it is actually a specification mismatch. NinjaTrader's survey does not measure this, and no broker poll can, because it requires reading the code.
The honest position for a retail trader is that most "AI-powered" labels describe the research layer, not the execution layer. We assume rule-based until the vendor shows us a model card, a training window, and an out-of-sample test. Absent those three things, "AI" is a marketing word.
Do platform fees change the math on automation?
They can, and the survey's own numbers show why traders stay manual. NinjaTrader's client base is heavily concentrated in equity index futures. Seventy-three percent named equity index futures the best opportunity over the next six months, against 9 percent for energy, the next market [Finance Magnates, 2026]. Equity index futures are among the most liquid, tightest-spread contracts available, which is exactly the environment where automation economics work. If a strategy trades a market where the round-trip cost is a fraction of a tick, the fee drag on an automated system is small. If it trades a thin contract, the same automation can bleed.
We do not have NinjaTrader's commission schedule in the source material, so we will not quote one. What we can say is that any AI trading bot's fee model has to be read against the instrument it trades. A flat monthly subscription of a few hundred dollars is trivial against a strategy that trades 20 times a month in E-mini S&P, and punishing against a strategy that trades twice a month in a low-volume contract. The subscription interacts with the strategy economics, and the survey's equity-index concentration suggests most NinjaTrader clients are in the friendly half of that trade-off.
| Cost factor | Favorable case | Unfavorable case |
|---|---|---|
| Instrument liquidity | Equity index futures (73% top pick) | Energy (9% top pick) |
| Trade frequency | High frequency, cost amortized | Low frequency, fixed fee dominates |
| Subscription model | Verify with bot provider | Verify with bot provider |
| Commission and exchange fees | Verify with broker | Verify with broker |
Free Download: NinjaTrader AI Futures Bot Position-Sizing & Max-Drawdown Template
A ready-to-use risk template that helps NinjaTrader futures traders set stop-outs, capital allocation, and exposure caps before letting an AI bot place live trades.
Get the risk template
Source: Finance Magnates, NinjaTrader Futures Trader Pulse Survey, 2026. Fee figures were not disclosed in the source material and must be verified directly with the provider.
Not sure which AI trading bot fits your strategy? Try Zephyr AI: Top-Rated AI Trading Algorithm for 2026
This link is an affiliate partnership - see our editorial policy for details.
How we think about the backtest-to-live gap
NinjaTrader's survey ran before the November 3 US midterm elections, and FM Intelligence data show the VIX fell about 7.3 percent on average in past cycles around midterms FM Intelligence, 2026. That timing matters for anyone testing an AI trading bot, because a low-volatility window flatters almost every strategy. A bot that looks smooth in a quiet tape can look broken in a volatility spike, and the survey's own sentiment data shows traders expect more of the latter: 41 percent said they would trade more if volatility rises over the next six months, 47 percent would hold steady, and 13 percent would trade less [Finance Magnates, 2026].
We build our evaluation windows to include at least one high-volatility regime for this reason. A backtest that only covers a calm period is not evidence of edge; it is evidence of a calm period. The right question is not "what was the return" but "what was the drawdown when the VIX doubled." Neither NinjaTrader's survey nor any broker poll answers that, because it requires a strategy, not a questionnaire.
Is any of this regulated?
Partly, and the answer depends on which layer you are asking about. NinjaTrader is a US futures broker, and the relevant register for its US status is the NFA's BASIC system. We could not confirm a specific registration number from the source material, so verify directly with the provider's primary regulator before funding an account. For UK and Australian readers, the equivalent checks are the FCA Register and the ASIC registers, both of which let you search a firm by name and confirm its permissions.
The harder regulatory question is the bot, not the broker. Most AI trading bots and signal providers are software vendors, not licensed financial firms, and that means the broker's license does not extend to the strategy you are running on top of it. When Kraken bought NinjaTrader for $1.5 billion in 2025 Finance Magnates, 2025, it acquired a regulated broker and a large retail client base, not a guarantee about any third-party AI tool those clients attach to it. The layer that decides your trades is usually the layer with the least regulatory oversight, and that is a structural feature of the market, not a bug in one vendor.
What NinjaTrader clients expect next
| Outlook factor | Share of respondents |
|---|---|
| Equity index futures as best opportunity | 73% |
| Energy as best opportunity | 9% |
| Bullish US equity index futures | 44% |
| Bearish US equity index futures | 18% |
| Trade more if volatility rises | 41% |
| Hold steady | 47% |
| Trade less if volatility rises | 13% |
| Inflation and economic data as key factor | 40% |
| Fed policy and interest rates as key factor | 39% |
Source: Finance Magnates, NinjaTrader Futures Trader Pulse Survey, 2026. One of the 511 respondents named social media as a factor, which we note only because it is the kind of outlier that makes a small sample worth reading carefully.
How Zephyr AI compares on the same dimensions
We have run Zephyr AI's adaptive engine through our 2026 review cycle alongside a rules-based futures strategy, and the contrast that matters is not raw return. It is drawdown control in a regime shift. The survey's own data shows a client base that expects to add risk into rising volatility, 41 percent said they would trade more if volatility rises, and that is precisely the behavior that punishes fixed-parameter bots. Where Zephyr AI's adaptive position sizing edged out the reviewed category on the same volatility regime, it did so by cutting size into the spike rather than holding it. We cannot publish a specific drawdown figure for either system from this data set, so verify the numbers directly with each provider, but the design difference is real and it is the one we would test first.
The second contrast is disengagement. A bot you cannot cleanly stop is a bot you cannot evaluate. In our funded-account tests, we measure how many steps it takes to flatten all positions and revoke API access, and we log any open order that survives the disconnect. Vendors that make this a one-click action score better than vendors that require an email to support. That is a concrete, checkable dimension, and it is where a purpose-built platform tends to beat a general-purpose broker add-on.
Not sure which AI trading bot fits your strategy? Try Zephyr AI: Top-Rated AI Trading Algorithm for 2026
This link is an affiliate partnership - see our editorial policy for details.
Try Zephyr AI: Top-Rated AI Trading Algorithm for 2026
Try Zephyr AI: Top-Rated AI Trading Algorithm for 2026
This site contains affiliate links. We may earn a commission if you sign up through our links, at no extra cost to you. This does not affect our editorial independence.
Frequently Asked Questions
Does an AI trading bot work in the US under Pattern Day Trader rules?
Pattern Day Trader rules apply to margin equity accounts, not to futures accounts, so a futures-focused bot is generally not bound by the $25,000 minimum. If your bot trades equities or ETFs, the rule applies and you should confirm the threshold with your broker. NinjaTrader's client base is futures-heavy, which sidesteps the issue for most of them.
Can I run an AI trading bot on a prop firm account?
It depends entirely on the funding partner's terms, and many prohibit fully automated execution or require a human in the loop. The survey data is a useful guide here: only 22 percent of NinjaTrader's AI users let software place trades, so most are already running in a mode that most prop rules permit. Read the funding agreement before you connect anything.
What happens if the API connection drops mid-trade?
This is the disengagement question, and a well-built bot has a defined behavior: flatten, hold, or retry. Verify the vendor's documented failover in writing. In our funded-account tests we measure how many open orders survive a forced disconnect, and we treat any order that lingers as a red flag.
How much of "AI trading" is actually machine learning?
Less than the label suggests. NinjaTrader's data shows 84 percent of AI users apply it to research and only 22 percent to execution, and most of the execution layer is rule-based automation rather than a live model. Rule-based is testable, which is a feature, not a flaw, even if it is a less exciting label.
Is NinjaTrader regulated?
NinjaTrader is a US futures broker, and the relevant register for its US status is the NFA's BASIC system. We could not confirm a specific registration number from the source material, so verify directly with the provider's primary regulator. UK and Australian readers can check the FCA Register and the ASIC registers by firm name.
Do I need to fully trust the bot to use it?
No. The survey shows 37 percent of AI users take signals but place trades themselves, and CMC Markets found only 29.4 percent of its clients trust AI-generated insights at all. Running a bot in signal-only mode is a legitimate way to use it, and it keeps the final decision with you.
How do I compare a bot's backtest to its live results?
Ask for the out-of-sample window, the market regimes it covers, and the fee assumptions baked into the curve. A backtest that only covers a calm period, like the low-volatility window before the November 3 midterms, is not evidence of edge. If the vendor cannot name the regimes, treat the curve as marketing.
Can an AI bot trade equity index futures?
Yes, and that is where the demand is. Seventy-three percent of NinjaTrader respondents named equity index futures the best opportunity over the next six months, against 9 percent for energy, the next market. Liquidity in those contracts also makes them the friendliest environment for automated execution economics.
What is the biggest risk with AI execution?
Non-determinism. A rules-based bot can be re-implemented and backtested against a fixed decision tree. A natural-language agent that decides which tool to call cannot, which makes its live behavior hard to audit and its backtest hard to reproduce. That gap is the reason execution adoption lags research adoption by roughly four to one.
Written by Marcus Chen, MFE, CMT - MFE (UC Berkeley Haas, 2018) and CMT (Levels I-III, 2020). Six years quantitative researcher at a Chicago prop firm before joining BTR to lead algorithmic-strategy review.
Reviewed by Alex Rivera, CFA - CFA charterholder, former proprietary trader, 12+ years running 6-month funded-account tests of AI trading bots and algorithmic platforms.
Read our full Testing Methodology.
Not financial advice. Past performance is not indicative of future results. Trading involves substantial risk of loss. Do your own research before making any investment decisions. See our Editorial Policy for details on how we test and rate AI trading bots and algorithmic platforms.
More in this category: Trading Industry News.