Clean Data or Costly Mistakes

Throughout my career working across operations, trading, portfolio management, and tax-loss harvesting platform development, one principle has remained constant: Data quality is foundational.

Not important. Not “nice to have.” Foundational.

A phrase I carried with me from my quantitative management years still applies today: “Garbage In, Garbage Out.” Nowhere is this more evident than in tax-aware portfolio strategies, where flawed inputs can materially erode tax alpha.

At Alphathena, we integrate with a wide range of data providers and custodians. One reality is clear. No dataset is identical. Each comes with its own nuances, quirks, and failure modes. Financial data is rarely pristine. It’s shaped by multiple systems, interpretations, and operational processes before it ever reaches a platform.

Where Data Actually Comes From

To understand why data quality remains one of the hardest problems in portfolio technology, you need to understand the supply chain. It’s more complex than most people realize.

Custodian data doesn’t start at the custodian. It flows through a layered ecosystem: clearinghouses like DTCC, pricing vendors, market data providers, and internal custodian accounting systems.

By the time it reaches your portfolio platform, it has passed through multiple transformations, validations, and formatting standards. Each one introduces potential inconsistencies.

Each custodian has developed its own data architecture over decades, shaped by system upgrades, acquisitions, regulatory demands, and layers of operational refinement.

The result is predictable but often underestimated.

The same portfolio can appear materially different depending on whether the data is sourced from Schwab, Fidelity, or Interactive Brokers. Field names vary. Asset classifications don’t always align. Cost basis methodologies differ. Even the timing of data updates can fall out of sync.

Then there’s delivery. Custodians and vendors distribute information through a wide range of channels, from Secure File Transfer Protocol to cloud storage like S3 buckets to real-time APIs.

At Alphathena, our developers built an architecture to ingest, reconcile, and normalize these varied formats so we can operate seamlessly across complex data ecosystems.

Corporate Actions: Where Things Get Interesting

Now let’s add another variable to the equation: corporate actions.

When Netflix executed its 10-for-1 stock split on November 17th, 2025, each custodian processed and published the update on its own timeline.

Before the split, an investor holding 100 shares at $1,000 per share had a position worth $100,000. After the split, they should have held 1,000 shares at $100 per share. Same $100,000 market value. Simple math.

Except it wasn’t simple.

For a window of time, the same security appeared drastically different depending on the dataset being referenced. Some custodians updated the price but not the shares, creating a phantom loss for days after the split.

Day 1 (Split Date)

Had these discrepancies gone unnoticed, a tax-loss harvesting strategy could have acted on outdated pricing and position data, potentially harvesting a loss that didn’t truly exist.

In automated portfolio environments, these aren’t minor errors. They are trade-triggering events.

At Alphathena, we systematically flag these discrepancies and place a global trade restriction for a couple of days until the data settles. Prevention beats correction every time.

Account-Level Issues Are Just as Dangerous

Data challenges aren’t limited to individual securities. Account-level problems can be equally disruptive, especially in large multi-custodian environments.

It is not uncommon for inactive accounts, closed accounts, or duplicate accounts to continue feeding into data streams longer than expected. These ghost accounts can distort cash calculations and lead to users inadvertently including them in trading workflows.

At Alphathena, we implement alerts when anomalies are detected. These accounts get flagged for users, ensuring they are properly excluded from trading. We provide transparency so users can track which active accounts are eligible to trade on the platform.

Why This Matters More Than Ever

Data quality is often viewed as an operational concern. Something that happens behind the scenes. The reality is that every portfolio outcome depends on it. As portfolio technology becomes more automated and tax strategies grow more sophisticated, the margin for error narrows. Small inconsistencies upstream cascade into real trades with real tax consequences.

The platforms that succeed in this environment won’t be the ones with the most features. They’ll be the ones engineered to detect anomalies early, before they become costly mistakes.