WhitmanTrading

Goldman Sachs: A Forecast Is Not a Trade

Goldman Sachs is an investment bank: it advises on transactions, underwrites issues, trades, and manages assets. Its published forecasts circulate widely as news, but they are house views produced for institutional clients and carry no risk rule a trader could act on.

How it works

A candlestick chart of the site's shared price history. The headline on the chart reads: An investment bank with a very long shadow.
An investment bank with a very long shadow. Illustrative chart - not real market data.

It is an investment bank, and everything follows from that. The recognisability is unusual; the business is not, and the general page on the model explains most of what this firm does.

A gently rising stretch of the long price series with an account equity curve beneath it. The headline on the chart reads: Advisory, trading, asset management and now consumers.
Advisory, trading, asset management and now consumers. Illustrative chart - not real market data.

Four lines of business. Advising on transactions and underwriting issues. Trading and market-making. Managing assets for institutions and wealthy individuals. And, more recently, consumer-facing products — the newest and smallest part.

A calmly advancing stretch of the long price series with a slowly rising equity curve beneath it. The headline on the chart reads: Its earnings follow deal volume and market activity.
Its earnings follow deal volume and market activity. Illustrative chart - not real market data.

The revenue is cyclical in a specific way. Cheap borrowing produces mergers and flotations, and volatility produces trading revenue. Neither tracks the stock market directly, which is why the shares behave differently from a broad financial index.

A flat, quiet stretch of the long price series with a gradually rising equity curve beneath it. The headline on the chart reads: Its clients are institutions, not people with accounts.
Its clients are institutions, not people with accounts. Illustrative chart - not real market data.

Nothing in the core business is aimed at a retail trader. Recognising that removes most of the confusion about what its published output is for.

What the forecasts actually are

A strongly rising stretch of the long price series with an account curve breaching its limit. The headline on the chart reads: Its published forecasts are news, not instructions.
Its published forecasts are news, not instructions. Illustrative chart - not real market data.

A house view is published on a schedule and revised as conditions change. It circulates as a headline because the name carries weight, and by the time it reaches a general audience the institutional clients have had it for some time.

A choppy, directionless stretch of the long price series. The headline on the chart reads: And forecasts from any bank have an ordinary hit rate.
And forecasts from any bank have an ordinary hit rate. Illustrative chart - not real market data.

Forecasting is hard for everybody, including the people who are good at everything else. The base rate for macro and price forecasts is unimpressive across the industry, and a strong brand does not change the underlying difficulty of predicting.

A declining stretch of the long price series. The headline on the chart reads: None of this gives a retail trader an edge.
None of this gives a retail trader an edge. Illustrative chart - not real market data.

A view that has been on the newswires is priced. Whatever information content it had was expressed in the market before it reached you, which is the general problem with acting on published research.

In practice

A 72-bar candlestick section of the shared price history with an account curve shown with and without fees. The headline on the chart reads: The fees are taken before the outcome is known.
The fees are taken before the outcome is known. Illustrative chart - not real market data.

The firm is paid for arranging, not for being right. That is the structure of the business rather than a criticism, and it explains why the research and the advisory sides need separating internally.

A candlestick chart with a volume histogram beneath it, with the volume histogram emphasised. The headline on the chart reads: Its desk may be the other side of your trade.
Its desk may be the other side of your trade. Illustrative chart - not real market data.

Market-making means quoting both sides and managing inventory. There is no adversarial intent in that, and there is also no duty to your outcome.

A long-horizon candlestick view of the same price series. The headline on the chart reads: The business is cyclical over years.
The business is cyclical over years. Illustrative chart - not real market data.

The cycle runs in years, not quarters. Deal droughts and deal booms both persist, so a single quarter’s results say little about the trajectory.

A candlestick series containing several opening gaps, with the largest opening gap marked. The headline on the chart reads: And its target changes move prices before you see them.
And its target changes move prices before you see them. Illustrative chart - not real market data.

By the time a rating change is reported, the opening gap has happened. The information reached clients first, and the price reflects it before the article exists.

A declining stretch of the long price series, with the entry price and the level at which a stop would trigger drawn as horizontal lines. The headline on the chart reads: A price target is not a stop and carries no risk rule.
A price target is not a stop and carries no risk rule. Illustrative chart - not real market data.

A target has no invalidation attached. It tells you a destination and never tells you where the idea is wrong, which makes it unusable for a stop and therefore unusable for sizing.

A candlestick chart of the site's shared price history, annotated with the round-trip cost. The headline on the chart reads: Every round trip costs 2% of a bar.
Every round trip costs 2% of a bar. Illustrative chart - not real market data.

Acting on a headline costs the same as acting on anything else. A round trip on this site’s shared history is 2% of a median bar’s range, and a rating change is rarely worth more than that by itself.

Using bank research the way it can be used

The forecast is the least valuable part and the modelling is the most. A research note contains industry structure, capacity data, contract terms and cost assumptions gathered by people who spend their working lives on one sector.

Read the assumptions and ignore the conclusion. You can disagree with an assumed growth rate and see what happens to the answer; you can do nothing at all with a rating. A note that states its assumptions clearly is worth reading even where you think it is wrong, and one that hides them is worth nothing whatever it says.

And treat the target as a piece of market information rather than a piece of analysis. Knowing what a widely followed firm has published tells you something about what other participants have read, which is occasionally useful and is not the same as knowing what a company is worth.

What Goldman Sachs is not

It is not a retail broker. The clients are institutions.

Its research is not advice to you. It is a house view.

A target is not a plan. There is no invalidation.

And its forecasts are not better than the base rate. Forecasting is hard.

When it fails

A sideways, range-bound candlestick series. The headline on the chart reads: In a quiet market its trading revenue falls away.
In a quiet market its trading revenue falls away. Illustrative chart - not real market data.

In a quiet market both engines slow at once. Deals stop happening and trading volumes fall, so the revenue is exposed to conditions in a way a fee-on-assets business is not.

The second failure, for a reader, is treating a published view as an edge. It was priced before publication.

A third is trading a target with no invalidation. There is nowhere to be wrong, so there is no size.

A fourth is inferring the firm’s positioning from its research. The desks are separated deliberately.

A fifth is assuming the brand improves the forecast. It improves the distribution.

And a sixth is reacting to the headline version. The gap on the chart already happened.

The original data

Of the 24,971 videos in research/search-study-corpus.jsonl, 2 have “goldman sachs” in the title, and one of them has 5,610,812 views. Investment banks appear in 16 titles at a median of 88,219 and hedge funds in 22 at 83,975. The counts are in research/broker-coverage.json.

A strongly rising stretch of the long price series, cut short at the decision bar. The headline on the chart reads: The bank just raised its target. Follow it?
The bank just raised its target. Follow it? Illustrative chart - not real market data.

Two videos, and one of them was watched five and a half million times. That single upload is among the most-viewed things in this entire corpus, and its framing is worth noticing — it is an ex-employee talking, not an explanation of what the firm does. The name sells the video; the business does not.

A note on that figure: the raw corpus file listed this video twice, which is why the deduplicated count is two videos rather than three. It is the reason every count on this site is now taken from unique video identifiers rather than rows.

The answer to that final question is no, and specifically because you cannot size it. A target with no invalidation gives you nothing to put a stop against, and a position without a stop has no defensible size. Read the assumptions behind the target and form your own view of them — that content is genuinely valuable, and the number on the front page is not.

Investment bank is the business model this firm is an instance of. Market makers covers the trading half and why the quote is an inventory decision. And hedge fund is the client on the other side of much of this activity.

What I actually do

I used to read the headline versions of these forecasts as though somebody had settled the question. What changed my mind was noticing that the number always came without the thing I actually needed — what would have to happen for it to be wrong. A target without an invalidation is not something you can trade, however good the firm producing it.

— Michael Whitman

This page is educational, not financial advice. Test every idea on your own charts before risking money.