Between late 1982 and early 1983, the California firm Delphi Associates carried out one of the most unusual documented civilian applications of remote viewing: nine consecutive silver-futures trades on the COMEX, all nine predicted by a precise protocol, all nine correct, with a documented profit of around USD 120,000 for the participating investors. The method that made this possible – associative remote viewing (ARV) – is conceptually interesting enough that it is still discussed and replicated in academic and semi-academic literature nearly forty years later. This article reconstructs the experiment step by step, describes the subsequent failure on scaling, and places the lesson in context.
Who was Delphi Associates?
In the autumn of 1982, Russell Targ left the Stanford Research Institute, where he had spent ten years building the remote-viewing programme together with Harold Puthoff (see the Puthoff/Targ portrait and the SRI Geller experiments article). Together with the psychologist and sensitive Keith Harary – an experienced viewer himself – he founded Delphi Associates in California, a small consulting and applications firm with the explicit aim of transferring the methodology developed at SRI into civilian fields. The principal funder and investor of the silver-futures programme was Paul Temple, a Texas businessman and long-time private patron of psi research.
The constellation was unusual: two scientifically trained practitioners (Targ as physicist, Harary as psychologist), an investor with real risk capital, and a market that permitted an objective, financially verifiable evaluation. Unlike session-based parapsychological tests, the outcome was not a statistical hit rate in a laboratory but the question whether, at the end of the trade, money was actually there.
The basic problem of direct psi market prediction
Anyone who, in a remote-viewing setting, tries to see a market directly – for instance, "describe how the silver price closes tomorrow" – runs into a series of systematic problems. The market is abstract; it has no sensory shape that a viewer could grasp. Numbers, charts and trends are secondary constructs, not original objects of perception. And the viewer's own wanting – to make money – is a direct trigger of psi missing: as soon as the task becomes instrumentally alienated, the hit rate collapses. This problem has been well known since the earliest tests in the Wigner-adjacent configurations and the historical mediumship records from William Crookes to Charles Richet.
Targ and Harary's innovation was a methodological trick that lets the viewer not look at the market at all.
Associative remote viewing (ARV) – the protocol in detail
The basic idea: the viewer is asked not to predict the market movement, but to describe a future sensory perception event. Which perceptual event the viewer will experience depends on the market outcome – but the viewer does not know that.
Step by step through a trial
- Selection of the two target objects (day T, before trial): an experimenter not involved in the viewing chooses two clearly different physical objects – for instance a rough piece of cork bark and a smooth piece of cloth, a feather and a stone, a plush teddy bear and a metal hammer. One object is assigned to upward market movement, the other to downward. Which represents what is recorded and sealed; the viewer is told nothing.
- The viewing (day T): Harary is asked to direct his attention to a future, precisely defined perceptual event – for example, "the object that you will hold in your hand the following day at 4 pm". He describes freely what he "sees": shapes, textures, colours, sizes, temperatures.
- Judge comparison (day T): a third party, also not involved in the viewing, receives Harary's description and the two target objects and decides which of them the description matches more closely. This sets the market prediction.
- Trade (day T+1, before market open): Paul Temple places the corresponding silver-futures trade on the COMEX. If the judge identifies the "upward object", they go long; if the "downward object", short.
- Market resolution (day T+1, market close): the actual day's close is determined. The corresponding market movement is compared with the sealed assignment key, and this defines which of the two objects was the "real target" of this trial.
- Feedback (day T+1, after market close): Harary is shown that single object – and only that one – physically, given it in his hand, asked to describe it. This closes the perceptual loop: his viewing the previous day is in fact tied back to a concrete sensory experience.
The decisive point: Harary at no time sees the non-matching object. This rules out a number of conventional sources of error:
- He cannot see the market directly and therefore cannot "miss" it.
- He cannot build a learning effect on "wrong" feedback that would skew later trials.
- He does not know which market outcome is assigned to which object – the assignment is sealed.
- His own financial interest does not couple directly to the task; the task is phenomenological, not commercial.
ARV is thus an elegant example of how a good protocol makes psi effects visible by removing from the viewer's immediate perceptual space the very things that typically destroy them (pressure, instrumental self-interest, wrong feedback).
Phase 1 – the nine-of-nine series
The first phase ran for about nine weeks, from late 1982 into the spring of 1983. One to two trades per week. The result is consistently documented across several sources: nine consecutive trades, all nine correct. The chance probability for nine correct binary predictions in a row is 1 in 512 – less than 0.2 percent. The cumulative profit from the corresponding COMEX trades was around USD 120,000 in then-current dollars (inflation-adjusted for 2026 this is roughly USD 400,000).
With this series, the case was picked up in the financial press. The Wall Street Journal ran an extensive report on the experiment shortly afterwards, and over the following years the case was treated in several investment magazines, talk shows and a large number of books on applied parapsychology. Targ and Harary published their own account in 1984 in the book The Mind Race (Villard / Random House) – a primary source written one year after the experiment.
Phase 2 – the failure on scaling
After the success of the first series, the Delphi team tried to extend the procedure – more viewers, more trades, larger investment volume, more involved persons. The hit rate collapsed. In some accounts, the later trials fell markedly below chance expectation – a classic psi-missing effect that has been regularly observed in parapsychological research since J. B. Rhine in the 1930s.
Targ and Harary have since discussed this collapse extensively and attribute it to several interlocking factors:
- Group dynamics: more participants meant more inconsistent expectations, more conflict over trade decisions and more emotional charge in individual sessions.
- Greed and pressure: as the financial stakes grew, the sessions were experienced differently – more anxious, more focused on the desired outcome, less of the relaxed openness that had marked the first phase.
- Sceptics in the room: as the operation scaled, participants joined whose methodological scepticism was warranted, but whose scepticism acted as a psi inhibitor in the session – an effect that William Crookes with Florence Cook and Charles Richet with Eusapia Palladino had already described.
- Loss of clean protocol: in practice, steps were shortened, feedback loops modified, the ARV structure softened.
The experiment was finally wound up in the course of 1983/84. Delphi Associates remained as a consulting firm for some further time, but gradually moved on to other research and training questions.
Methodological assessment
What makes the Delphi experiment scientifically so interesting is the double lesson:
What it shows
Phase 1, assessed methodologically, is a well-documented finding. Nine of nine correct binary predictions under a protocol that excludes the most obvious conventional explanatory routes:
- No lip reading, no verbal cue – the viewer does not know the object-to-market-movement assignment.
- No insider knowledge – object selection is independent of market information.
- No selective evaluation – every single trade was financially executed; one could not retroactively select the "good" trials.
- No subjective hit evaluation – an independent judge with no market knowledge evaluated the match.
What it does not show
Nine trials are, statistically, a small series. A 9-of-9 result at p = 1/512 is significant, but not enough to bear a broad hypothesis like "ARV works in general". What it shows is: under very specific conditions, nine times in a row, an effect was present that cannot be explained by chance. Phase 2 conversely shows: these conditions are not arbitrarily scalable.
How the participants see it today
Russell Targ has revisited the experiment in several later books – Limitless Mind (2004), The Reality of ESP (2012) – from the distance of several decades. His self-criticism is consistent: phase 1 was a real finding, but phase 2 showed him that psi effects are sensitive under purely instrumental–financial conditions. "If you only want to use psi to make money, it stops working" – a sentence that recurs in Targ's later talks.
Keith Harary has since partly withdrawn from academic psi research and reoriented to other fields of applied psychology. Paul Temple has also withdrawn from the field. Of the three, Targ alone is, today over ninety, still publicly active and continues to speak about the case.
Consequences: PEAR and the Applied Precognition Project
The Delphi protocol was methodologically convincing enough that it has been replicated repeatedly. The PEAR lab at Princeton ran a series of ARV studies in the late 1980s and 1990s, although without reproducing the spectacular financial results of the first Delphi series – usually with hit rates just above chance, but statistically significant across many hundreds of trials.
In the 2000s and 2010s the Applied Precognition Project (APP) emerged, founded among others by the horse-racing statistician Marty Rosenblatt, which systematically applies ARV to financial-market and sports-betting tasks. APP has over years documented statistically significant but smaller effects, and thereby confirms Targ's phase-2 lesson: ARV reliably delivers above-chance, but the hit rate of the spectacular phase-1 series is not sustainable.
What remains
The Delphi Associates silver-futures series is one of the few financially verified applications of a parapsychological method: not a laboratory effect with p-values, but real COMEX trades with real dollar amounts. It is methodologically secured by the ARV protocol in a way that makes conventional explanations difficult. And it failed on scaling in its phase 2 – in a way that does not invalidate the phase-1 observation but very clearly shows under what conditions it does not work.
Anyone engaging with the practical application of remote viewing will not be able to bypass this case – neither as evidence for the possibility nor as a lesson on its limits.
Sources: Russell Targ & Keith Harary, The Mind Race: Understanding and Using Psychic Abilities, Villard / Random House 1984 (primary source from the two participants). Jim Schnabel, Remote Viewers: The Secret History of America's Psychic Spies, Dell 1997 (chapter on Delphi Associates, based on interviews with all participants). Russell Targ, Limitless Mind, New World Library 2004, and The Reality of ESP, Quest Books 2012 (retrospective with self-criticism). Lance Mungia, Third Eye Spies, documentary 2017 (Targ recounts the experiment in his own words). Reporting in the Wall Street Journal 1984. PEAR replication studies in the 1990s, Journal of Scientific Exploration. Marty Rosenblatt et al., Applied Precognition Project (online, ongoing from around 2007).