Blogfoundation
Sugra Finance: the price does not say who is on the other side
A stock falls ten percent: panic or profit-taking? The number is identical either way. Holdings, entry price, disclosed shorts, insider flow and positioning against five years of its own history are what separate them, and the fundamentals are built from the filings rather than bought.
A stock falls ten percent. Is that panic or profit-taking?
The price cannot tell you, and this is not a philosophical point. The number is identical in both cases. What separates them is who is on the other side of the trade and what they paid, and that information exists, is published by regulators, and almost never sits next to the quote.
Take a name trading at 333.02 today. Institutions filing their positions were buying it through the first quarter of 2024 at an average of 181.83. Whatever a ten percent drop is for them, it is not a loss. They are eighty percent up and a bad week does not change a thesis. Now suppose that same average had been 320. Same price, same drop, entirely different set of people deciding what to do on Monday.
Quotes are commodity. Every vendor has them. The reason this direction exists is the layer underneath.
Six things the price does not say
Who holds it. For that name, 10,260 institutional filers, 10.85 billion shares, 2.59 trillion dollars of disclosed value in one quarter’s filings.
What they paid. Entry is reconstructed quarter by quarter, each with the number of trading days in the sample and a flag for whether the quarter is complete:
GET /api/v1/sec/13f/consensus/037833100/avg-entry-price?ticker=AAPL
2024 Q1 avg close 181.83 61 sample days
2024 Q2 avg close 186.49 63
2024 Q3 avg close 223.32 64
2024 Q4 avg close 235.68 64
...It is an approximation and the response says so by exposing its own construction: quarter boundaries, day counts, availability. You can argue with the method because the method is visible.
Who is short, on the record. Not an estimate of short interest but the regulatory disclosure itself, holder by holder above the reporting threshold:
GET /api/v1/fca/shorts/most-shorted
Vistry Group PLC 16.74% disclosed short 18 holders as of 2026-07-09
IBSTOCK PLC 13 holders
WH SMITH PLC 11 holdersEighteen separate funds disclosing against one issuer is a different fact from a single aggregate percentage, because it tells you whether a crowded short is one conviction or eighteen.
What insiders are doing with their own money. Monthly net purchase ratios built from the statutory forms:
GET /api/v1/insiders/sentiment?symbol=AAPL
2026-06 buys 0 sells 116 shares net ratio -100% 1 distinct insiderNote how small that is. One insider, 116 shares. The value of the field is not the headline number, it is that you can see the sample size and refuse to build a story on one transaction. A vendor sentiment score would have returned “negative” and told you nothing about how thin it was.
How futures participants are positioned, against their own history. This one is the most underused:
GET /api/v1/cot/extremes
NASDAQ MINI net position -74,690
percentile rank 0.2 over 261 weeks
4-week change -23,628 13-week change -35,690A net short of seventy-four thousand contracts means nothing by itself. In the 0.2nd percentile of five years of its own history, it means positioning is at an extreme it has almost never reached. The index can be flat while this moves to the edge of its range, and only one of those two facts shows up on a chart.
Where money is being placed on outcomes. Prediction market prices, order books and open interest for named events, which is a probability quoted by people with capital at risk rather than a forecast.
We build the fundamentals, we do not buy them
Ask what financial concepts are available for one company and the answer is not a fixed menu:
GET /api/v1/fundamentals/AAPL/concepts
entity: Apple Inc. cik 0000320193
us-gaap taxonomy: 503 concepts
dei taxonomy: 2 conceptsFive hundred and three, because that is what the company actually reported under the accounting taxonomy, not the twenty fields a vendor decided were interesting. Pull any one of them and it arrives with its formal definition attached:
Revenues. Amount of revenue recognized from goods sold, services rendered, insurance premiums, or other activities that constitute an earning process. Includes, but is not limited to, investment and interest income before deduction of interest expense when recognized as a component of revenue…
That paragraph is the difference between a number and a number you can defend. When two sources disagree about “revenue”, they are usually reporting different concepts, and without the definition you cannot even locate the disagreement.
The same construction runs sideways across companies:
GET /api/v1/fundamentals/frames/Assets?period=CY2024Q4I
total_companies: 6,248 source: SEC EDGAR XBRL Frames
1 FEDERAL NATIONAL MORTGAGE ASSOCIATION 4,349,731,000,000 as of 2024-12-31
2 JPMorgan Chase & Co
3 Federal Home Loan Mortgage CorporationSix thousand two hundred and forty-eight filers reporting one concept for one period, each with its identifier and period end. That is not a screener with preset columns. It is the filing record itself, indexed so you can rank by any concept a company reported.
Because it is built from filings rather than bought as a package, the cuts that normally do not exist are simply there: debt maturity schedules, lease detail, tax detail, bank-specific metrics, share and equity-compensation structure, and full-text search across the filings themselves.
Computed on the platform, with the formula in the response
The second half of the direction is quantities that no upstream publishes, because somebody has to calculate them.
GET /api/v1/options/AAPL/implied-move
underlier_price 333.5362
expiry 2026-07-27 (3 days)
atm_iv 0.1869
implied_move_pct 1.6938
implied_move_dollar 5.6496
implied_range 327.8866 - 339.1858
iv_source cboe_native
methodology 1-sigma straddle-equivalent implied move:
implied_move_dollar =
underlier_price * atm_iv * sqrt(days_to_expiry / 365.25)
...Calendar-day year convention (365.25) matches the Greeks
calculation in the underlying snapshot. The range is
symmetric around the spot; real-world moves are
typically skewed.The options market is pricing a move of plus or minus 1.69 percent over three days, and the range is in dollars so it can go straight into a brief.
Read the methodology field again, because it does two things almost nobody does. It carries the actual equation, down to the year convention and why that convention was chosen. And its last sentence tells you the model is wrong in a specific direction: the range is symmetric, real moves are skewed. A number that ships with its own caveat is worth more than a better number that ships bare.
Alongside it sits a precomputed implied-volatility surface with term structure, strikes and moneyness, and 113 indicators evaluated on the platform, which means the common pattern of pulling years of history to your own machine just to compute a moving average stops being necessary.
One key, nine asset classes
Equities, options, ETFs, futures, foreign exchange, crypto with its on-chain and DeFi layers, fixed income, commodities, and prediction markets.
That list is normally five vendors, five contracts, five schemas, five invoices and five rate limits, plus the glue code that reconciles their symbologies. Here it is one header and one envelope. The claim is deliberately a list rather than an adjective, because a list can be checked.
The trap that would have wrecked my own example
Worth stating plainly because I nearly walked into it while writing the News piece.
The same commodity can carry more than one price here, at different resolutions, from different institutions. Crude is the clearest case:
monthly index 62.72 period 2025-12
daily series 86.99 date 2026-07-20Both are correct. One is a monthly benchmark index published with a long lag, the other is a daily official series. Twenty-four dollars apart and seven months apart, and nothing in the field name warns you.
Pair the daily series with anything time-sensitive and the analysis works. Pair the monthly index with a weather event or a shipping disruption and you will produce a confident chart in which cause follows effect by half a year. The resolution of your slowest leg sets the resolution of your conclusion, and choosing the wrong layer is not a small error, it is the whole error.
Same discipline applies to agricultural names: monthly benchmark indices and daily exchange contracts both exist, and only one of them can answer a question about this week.
A price without a regime is unreadable
A ten percent fall in a world where money costs three percent is a different event from the same fall where money costs six. Nobody disputes that, and almost nobody wires it into the product, because the two numbers live in different systems.
They do not have to. The official record and what the market paid to believe are both one call away, and the interesting part is where they disagree.
policy rate, upper bound 3.75 (2026-07-25)
government bond curve 1M 3.80 6M 4.08 2Y 4.33
3M 3.96 1Y 4.14 3Y 4.36The whole short end is priced above the policy rate and climbing. Whatever the official stance is, nobody transacting at those levels is positioned for it to loosen.
Now the same comparison on inflation. The published rate is 3.46 percent year over year. The inflation-protected curve says the market is paying for 1.99 percent over two years and 2.32 over five.
Together those are one statement, not three: rates stay high, and that is precisely why inflation is expected to fall. Hold a position against that backdrop and the question “is this drawdown a problem” has a different answer than it would if the curve were pricing cuts and rising prices.
This is why the Macro direction is not a separate product for a separate audience. It is the denominator under everything in this post.
And a price without physical confirmation is a guess about why
The other join answers a different question: not what conditions the move happened in, but whether the reason offered for it is real.
A price move raises a question the price cannot close. Is this repricing a risk, or repricing an actual disruption? Press coverage is compatible with both, and reading more of it does not separate them, because both worlds generate the same sentences.
Physical movement does separate them. When tanker tonnage through a strait collapses by a factor of fifty over the window in which oil gains twenty-seven percent, the market is paying for cargo that stopped, not for a fear that has not reached the water. Had transit held near its normal band with identical coverage, you would be looking at a premium that unwinds on a headline, because nothing physical needs repairing for it to disappear. Same news, opposite trade. The worked example is in News.
The rule that comes out of both joins: the price is the observation, the regime tells you what it means, positioning tells you who is exposed to it, and physical measurement tells you whether the stated reason is true.
Where this stops
None of this is advice, and several of the layers are slower than they look.
Institutional filings arrive quarterly and late, so holdings describe a position that existed at a snapshot date and may have been unwound before you read it. Entry price is reconstructed from quarterly averages, not from trade tapes: it is an approximation and should be treated as one. Disclosed shorts only exist above a regulatory threshold, in the jurisdictions that require disclosure, so an empty result means nothing was disclosed rather than nobody is short. Positioning reports publish weekly with their own lag.
We return market data and platform analytics. We are not a broker, we do not route orders, and there is no score in here that tells you what to do. If a number in your product implies an action, you wrote that part, and you own the formula and the disclaimer.
Running it
H="x-api-key: $SUGRA_API_KEY"
B=https://sugra.ai
# the observation
curl -sS -H "$H" "$B/api/v2/quotes/AAPL/price"
curl -sS -H "$H" "$B/api/v2/quotes/AAPL/historical?range=1mo"
# who holds it, and what they paid
curl -sS -H "$H" "$B/api/v1/sec/13f/consensus/037833100"
curl -sS -H "$H" "$B/api/v1/sec/13f/consensus/037833100/avg-entry-price?ticker=AAPL"
curl -sS -H "$H" "$B/api/v1/sec/13f/holders-of/037833100?limit=25"
# who is short, and what insiders are doing
curl -sS -H "$H" "$B/api/v1/fca/shorts/most-shorted"
curl -sS -H "$H" "$B/api/v1/insiders/sentiment?symbol=AAPL"
# fundamentals built from the filings themselves
curl -sS -H "$H" "$B/api/v1/fundamentals/AAPL/concepts"
curl -sS -H "$H" "$B/api/v1/fundamentals/AAPL/history/Revenues"
curl -sS -H "$H" "$B/api/v1/fundamentals/frames/Assets?period=CY2024Q4I&top_n=25"
# computed on the platform, formula included
curl -sS -H "$H" "$B/api/v1/options/AAPL/implied-move"
curl -sS -H "$H" "$B/api/v1/options/AAPL/iv-surface"
# positioning against its own five-year history
curl -sS -H "$H" "$B/api/v1/cot/extremes"
# the regime underneath the price
curl -sS -H "$H" "$B/api/v1/fred/series/DFEDTARU?limit=2"
curl -sS -H "$H" "$B/api/v2/fixed-income/treasury/yield-curve"
curl -sS -H "$H" "$B/api/v2/fixed-income/treasury/tips-curve"
# mind the resolution
curl -sS -H "$H" "$B/api/v1/commodities/energy/petroleum" # daily
curl -sS -H "$H" "$B/api/v1/commodities/prices" # monthly indexLog the entry price next to the quote. A drawdown against a cost basis is a fact about people; a drawdown against nothing is a fact about a chart.
Market data and platform analytics. Not investment advice. Values pulled live 2026-07-26.
Previous: Platform intro. Next: Sugra Macro. Full series: Platform intro.