algoseek

The de facto standard for professional market data and infrastructure. 20+ years of tick history, real-time feeds, co-location and cloud SQL.

[algoseek @AlgoseekData]

If you have studied US microstructure, you have probably touched our tape.

Equities from January 2007, OPRA, CME Globex. Every trade and quote, one file per symbol per day

We recorded it off the feed. Nothing here is inferred.

https://t.co/NWLOgmssMJ

[algoseek @AlgoseekData]

@P123Finance The key distinction is proof vs evidence. You may need many years of portfolio returns for statistical proof, but deterioration in cross-sectional IC, factor spreads, or implementation costs can provide evidence of decay much earlier.

[algoseek @AlgoseekData]

Since July the AI hardware complex is down ~22% and energy is up ~18%.

Energy is the best sector of the last two months, and over the full year the median energy stock is up 41% against 36% for chips.

44 of the 50 most traded US energy stocks are up since July

[algoseek @AlgoseekData]

@BullTheoryio That $163B number is highly conditional. Algorithmic selling doesn't trigger the drop; a massive decline must happen first to force their hands. Volume scales with the decline. We'd only see the full $163B if the market falls off a cliff, and it'd happen over multiple sessions.

[algoseek @AlgoseekData]

@RealJGBanks Hindsight bias can make any discretionary pattern look like a profitable system. Usually collapses when you try to build a strategy around it.

[algoseek @AlgoseekData]

@PeterMallouk Exactly, the short term may be driven by sentiment and headlines, but over time, earnings growth tends to do the heavy lifting. Patience is often the hardest part.

[algoseek @AlgoseekData]

In the most active US stocks, half of all best bids are gone in under a millisecond.

We timed every change to the best price in Apple, Microsoft and Tesla for a session. Median life: 0.69 ms.

The price on your screen is a description of the past.

[algoseek @AlgoseekData]

@MikeZaccardi Good distinction. If the term premium is flat while yields rise, that points more toward a repricing of the expected policy path than a sudden loss of fiscal confidence. Still, term-premium estimates are model-dependent, especially when comparing countries.

[algoseek @AlgoseekData]

The middle of the US trading day is drying up two and a half times faster than the close.

Since January, shares traded per session fell 26.5% in the midday and 10.6% in the final hour.

Now, where you trade in the day matters more than it did before.

[algoseek @AlgoseekData]

@sentimentrader The key question is the baseline win rate: how often is the S&P 500 higher two months later without this signal? If that’s already 65–70%, a 72% hit rate isn’t particularly strong evidence of an edge.

[algoseek @AlgoseekData]

@TaviCosta It's worth adding that the move came from shrinking participation, median daily volume in Newmont fell 36% from January to mid-May, Freeport 30%, Agnico 20% and so on.

[algoseek @AlgoseekData]

Open any strategy you have backtested.

How much did it lose on companies that went bust?

If the answer is barely anything, that's not skill,
It's the data stopping before the bad part.

https://t.co/c4EggeovLs

[algoseek @AlgoseekData]

More information👇https://t.co/NWLOgmssMJ

[algoseek @AlgoseekData]

A standard minute bar contains 5 numbers (OHLCV)

Ours carry 90.

So we tested whether the extra 85 numbers do anything.

Same model, same NVDA minutes, same target.
Out of sample data only.

[algoseek @AlgoseekData]

19 August was the biggest session of 2026 in the US long bond.

The T-bond future traded 997 623 contracts, the highest of any day this year.
2.3x larger than the median session.

The front contract gained 1.29% open to close, its largest up day of 2026.

[algoseek @AlgoseekData]

This is often a reason why a backtest performance doesn't hold up live trading

[algoseek @AlgoseekData]

Your minute bars hold prices no order could have reached

59% of one-minute bars in the US top 1,000 have a high or low set by a print nobody could trade. 90% on 2026 data

1.01 billion bars, measured on our own tick tape

So we backtested both versions
https://t.co/TV0P0sE687

[algoseek @AlgoseekData]

#futurestrading #fintech #data #trading #machinelearning #quantitativetrading #algorithmictrading

[algoseek @AlgoseekData]

New Product Alert!

Algoseek's Lease Asset Package is a one-stop shop for all your market data and services needs. Its benefits include access to all available history, ongoing daily updates, future product enhancement ... 1/2

[algoseek @AlgoseekData]

#algoseek #themarketdatacompany #datasets #marketdata #historicalmarketdata #intradaytrading #algotrading #intraday #minutebars #equity #equitytrading #options #optionstrading #futures