QuantConnect/Lean · error · AssertionError

The total number of insights should be {expected}. Actual: {

Error message

The total number of insights should be {expected}. Actual: {self.insights.total_count}

What it means

on_end_of_algorithm assertion in a framework regression using HistoricalReturnsAlphaModel. It pins the alpha model's emitted insight count to exactly 78. A different count means the alpha's insight-emission logic, the universe selection, or the consolidation schedule changed. This is a behavioral contract test for HistoricalReturnsAlphaModel.

Source

Thrown at Algorithm.Python/HistoricalReturnsAlphaModelFrameworkRegressionAlgorithm.py:30

# limitations under the License.

from AlgorithmImports import *
from BaseFrameworkRegressionAlgorithm import BaseFrameworkRegressionAlgorithm
from Alphas.HistoricalReturnsAlphaModel import HistoricalReturnsAlphaModel

### <summary>
### Regression algorithm to assert the behavior of <see cref="HistoricalReturnsAlphaModel"/>.
### </summary>
class HistoricalReturnsAlphaModelFrameworkRegressionAlgorithm(BaseFrameworkRegressionAlgorithm):

    def initialize(self):
        super().initialize()
        self.set_alpha(HistoricalReturnsAlphaModel())

    def on_end_of_algorithm(self):
        expected = 78
        if self.insights.total_count != expected:
            raise AssertionError(f"The total number of insights should be {expected}. Actual: {self.insights.total_count}")

View on GitHub (pinned to d2c3659f87)

Solutions

Diff HistoricalReturnsAlphaModel (Algorithm.Framework/Alphas) and BaseFrameworkRegressionAlgorithm for changes to rebalance frequency, insight magnitude/threshold, or universe.
Check the regression's set_universe / warm-up config; fewer data points yields fewer insights.
Recompute the expected count against current code and, if the new behavior is correct, update expected = 78 with a justification and re-baseline other affected regression stats.

Example fix

# before: alpha emitted a different number after changing rebalance
self.set_alpha(HistoricalReturnsAlphaModel(rebalance=Resolution.DAILY))
# after: keep the default rebalance the regression was baselined against
self.set_alpha(HistoricalReturnsAlphaModel())

Defensive patterns

Strategy: validation

Validate before calling

# after setting the alpha, log expected insight cadence for review
self.set_alpha(HistoricalReturnsAlphaModel())
# at end, fail with context if counts drift
if self.insights.total_count != 78:
    self.debug(f"insight count drift: {self.insights.total_count}")

Prevention

Pin alpha-model constructor arguments; do not silently change rebalance/resolution.
When updating the framework, recompute and document insight-count baselines.
Run the full framework regression suite after any Alpha/Universe change.

When it happens

Trigger: self.insights.total_count != 78 at algorithm end. The alpha emitted more or fewer insights than expected over the regression period.

Common situations: A change to HistoricalReturnsAlphaModel insight emission (frequency, threshold, rebalance); universe size change in BaseFrameworkRegressionAlgorithm; resolution/warm-up period change altering how many historical-return samples are available; consolidation or scheduled-event timing change.

Related errors

AI-assisted analysis of QuantConnect/Lean@d2c3659f87 (2026-08-13). Data as JSON: /api/errors/68a738ab79ede4b8. Report an issue: GitHub.