The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →In Michael Hairetis’s report on 118,857 three-second Nasdaq-100 futures bars, TimesFM 3.0 failed to beat a naive comparison in each of four tested approaches. That is a result for one five-session sample—not evidence that the model cannot forecast other markets or periods.
What the evaluation tested
Hairetis’s DEV Community article, posted October 1, 2026, describes five sessions of Nasdaq-100 futures data sampled in three-second bars, including bid/ask volume. The targets were midpoint price and an order-flow series described as cumulative delta minus volume signed by candle direction.
As an Amazon Associate I earn from qualifying purchases.
The article reports four approaches: direct forecasts, forecasts on one-minute bars, an anomaly-style residual made from the paired series, and a detector derived from forecast quantiles. Their results use different metrics, so they should not be collapsed into a single ranking.
Recommended Free Tools
Direct forecasts at a 36-bar horizon
For direct forecasting, the author reports 2,674 anchors and a 36-bar horizon—108 seconds at the stated three-second sampling interval. The reported errors were 7.4% worse than naive for both midpoint price and order flow. Directional accuracy was 50.0% for price and 54.4% for order flow.
#1 Best Overall
- Language: english
- Book - trading: technical analysis masterclass: master the financial markets
- It is made up of premium quality material.
“Worse than naive” describes the reported error comparison; directional accuracy is a separate measure of whether the forecast got the direction right. Neither figure, by itself, establishes whether a trading strategy would make or lose money.
Did one-minute bars help?
The author also aggregated the data into one-minute bars and evaluated a 36-minute horizon. The price result was 5.4% worse than naive, with 50.2% directional accuracy. The order-flow result was 19.2% worse than naive, with 44.9% directional accuracy.
Rank #2
- As a day trader, you can live and work anywhere in the world. You can decide when to work and when not to work.
- You only answer to yourself. That is the life of the successful day trader. Many people aspire to it, but very few succeed. Day trading is not gambling or an online poker game.
- To be successful at day trading you need the right tools and you need to be motivated, to work hard, and to persevere.
Coarser sampling therefore did not reverse the reported comparison. The two targets behaved differently on the stated metrics, and the percentages should be read separately rather than as interchangeable measures of performance.
Did reframing the task help?
Anomaly-style residual
Hairetis says the price and order-flow series mirror one another and describes summing their rolling z-scores to create a mean-reverting residual. That anomaly-style framing was still reported as 5.6% worse than naive by the error measure, with 51.6% directional accuracy.
Quantile-derived detector
The article says TimesFM’s forecasts supplied nine deciles. The author used those quantiles to estimate an up probability for each series and combine them into a detector. This test was not reported as a forecast-error comparison: agreement with the incumbent method was +0.000, and the highest-confidence group had a 24.7% hit rate against a 34.3% base rate.
Those detector figures indicate no reported advantage on the measures given. They do not show that the detector would perform the same way under different confidence-bucket definitions or market conditions.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How far the conclusion reaches
The results are a first-person evaluation of one model on five sessions of one futures instrument. The article’s available indexed text does not establish the session dates, exact naive-baseline implementation, market-data vendor, transaction costs or slippage, confidence-bucket construction, uncertainty intervals, or reproducibility artifacts. The author says the evaluation was causal and confined to sessions, but the available detail is insufficient to independently assess those safeguards.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteAccordingly, the defensible takeaway is narrow: in this reported sample, none of the four approaches improved on the naive comparison as measured in the article. The figures do not establish profitability, performance after trading costs, or broad failure across other instruments, periods, or regimes.
Source: Michael Hairetis, DEV Community, October 1, 2026. The available indexed article text is the basis for the reported figures; full methodology and supporting artifacts could not be verified.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
