Home of APS-MRCR · Stock-ranking research
Your investing idea.
Put to the test.
Meet APS-MRCR, TensorStreet’s neural model for ranking stocks. It learns from daily price paths, market context and monthly patterns. Build a portfolio around its signal and put the evidence to the test.
A research workspace for portfolios of US stocks. Explore ideas before deciding what deserves your capital.
An idea, made testable
Move the slider. See which stocks rise to the top.
2. Rank stocks using that rule
- 1Stock C7.6
- 2Stock A7.0
- 3Stock B6.6
- 4Stock D4.2
Fictional stocks and scores to explain ranking. These are not predictions or investment returns.
3. Turn the ranking into a portfolio.Buy the top 10%, rebalance each month, then test the history after trading costs.
No code needed. You’ll review every setting before a run.
Meet TensorStreet’s research model
Inside APS-MRCR
Multi-Resolution Convolutional-Recurrent network
Two stocks can finish a month with the same return after very different daily journeys. APS-MRCR learns from those journeys, their market context and the monthly picture.
A neural model built to rank stocks.
It is trained to order stocks by their next-month return. You turn that ranking into a portfolio, then test whether the signal adds useful information beyond a simple rule like momentum.
The output is a relative ranking score, not a price target or a probability of profit.
Three views of a stock. One signal.
Explore what each branch contributes.
60 trading days · 18 daily features
How did the stock get here?
A steady climb, a sharp reversal and a volatile recovery can end with the same return. This branch learns from the sequence of daily price moves, trading volume and technical features.
A CNN detects short patterns over 3 and 5 days. A GRU reads the sequence to learn how those patterns unfold over time.
Combine the views, accounting for uncertainty
Each branch produces a score and an uncertainty estimate. A learned gate combines them for each stock, reducing a branch’s influence when it reports greater uncertainty.
Three independently trained networks are averaged → one stock-ranking signal.
From the research manuscript · Table 1
Does combining the views help?
The combined model showed stronger agreement with subsequent stock rankings than either of these simpler versions in the paper’s test period.
| Model inputs | Rank IC |
|---|---|
| Daily prices only | 0.010 |
| Monthly inputs only | 0.024 |
| APS-MRCR · combined | 0.032 |
Rank IC measures agreement between the model’s ordering and realised returns. It is neither a return percentage nor a hit rate.
Learn from the past. Evaluate on later months.
In the research, the model is refit each January using earlier data. The most recent 12 training months select the stopping point; subsequent months are used for evaluation.
Historical study: Jan 2017–Jun 2026, using current S&P 500 members (survivorship bias); June is partial. These are manuscript results, not live performance or a forecast for your strategy.
For researchers: architecture, training and what remains to test
APS means AI Pattern Score. MRCR combines two CNN–GRU sequence encoders with a monthly MLP and an uncertainty-aware gate. The paper specifies 93,114 parameters and a three-seed ensemble. Daily inputs are observed through the previous month’s close; the target is the within-month rank of next-month returns.
The paper reports tests against momentum, linear and tree-based signals. Point-in-time index membership, an independent future-period replication, and a separate test of the uncertainty gate’s contribution remain open. The platform reuses committed historical predictions; opening a strategy does not train or freshly score the model.
Source: APS: Multi-Resolution Neural and Linear Price-Path Signals Beyond Momentum in Large U.S. Stocks, research manuscript, architecture and training section and Table 1. The reported comparisons describe that historical study; your portfolio settings can produce different results.
No account required. Runs when the research dataset is available.
A clear path from question to evidence
You don’t need a quant lab
to start thinking like a researcher.
TensorStreet brings the models and the testing tools together. You bring the idea. Start with a simple rule, then change one thing at a time.
- 01
Choose what matters
A signal is a rule that ranks stocks. Use APS-MRCR, recent returns, a classic indicator, or scores you upload.
- 02
Test the portfolio
A backtest applies your rules to historical data. Set the stocks, dates and trading costs, then see how the portfolio would have behaved.
- 03
Decide what to try next
Read the return, risk and caveats together. Compare with a simpler strategy, save the result, and refine your next question.
Start with curiosity
Which question would you like to answer?
Three starting points, with sensible settings already filled in. Pick one and make it your own.
Do recent winners keep winning?
Start with a simple idea: buy stocks with stronger past returns. A useful baseline for testing what APS-MRCR adds.
Explore this ideaTensorStreet’s research modelWhat can APS-MRCR see in price patterns?
Use TensorStreet’s neural stock-ranking model. It combines daily price paths, market context and 69 monthly features.
Explore this ideaTake the next research stepDoes APS-MRCR add to momentum?
Combine APS-MRCR with momentum. Compare the mix with either signal on its own to see what each contributes.
Explore this ideaGrow into your own workflow
Start in the builder.
Keep going in Python.
Learning the basics or already building models? The SDK uses the same engine as the visual builder, so you can move from a guided experiment to your own notebook.
Explore the Python workflowaps = ts.models.get("aps-mrcr")
mom = ts.indicators.momentum("12-1")
signal = 0.6 * aps.rank() + 0.4 * mom.rank()
bt = ts.backtest(signal, cost_bps=10)
bt.scorecard()A workspace that keeps your thinking
Leave with a result. Come back with a better question.
Your drafts, runs and saved scorecards stay together. Pick up where you left off, compare a new idea with your baseline, and build a research record you can revisit.
Do I need to know Python?
No. The visual builder lets you choose signals, set portfolio rules and run a historical test. Python is an optional next step.
Does TensorStreet place trades?
This release is for historical research. It does not connect to a broker or place orders. Live tracking is planned; historical results are not live returns.
What does a strategy mean here?
A repeatable set of rules for selecting and weighting a portfolio of stocks. A run is one historical test of those rules.
Can I trust a strong backtest?
A strong historical result is a starting point, not a guarantee. Current research data uses today’s S&P 500 members, which introduces survivorship bias. Scorecards show these limitations.