The transactional interpretation makes no experimental predictions different from standard quantum mechanics. But it does remove some of the puzzling and perhaps unnecessary assumptions that are part of other Interpretations of quantum mechanics. In particular, it denies that conscious observers are needed to cause the "collapse of the wave function" (without which there is no actual "outcome" in the measurement process).

The transactional interpretation adds nothing ad hoc to the standard theory, such as "hidden variables or additional terms to the Schrōdinger equation to force a collapse. It is explicitly indeterministic and non-local. Cramer has explored the radical possibility of sending information between entangled particles faster than the speed of light, as well as causal relations that go backwards in time (retrocausality). And, like Schrōdinger and the decoherence advocates, Cramer denies the existence of particles!

The core physics in the transactional interpretation is a way of looking at photon emissions and absorptions as an exchange of advanced and retarded waves that is based on the 1945 Wheeler-Feynman Absorber Theory of radiation, which was abandoned by Feynman, who went on to develop the Path Integral formulation of quantum mechanics and later, with Julian Schwinger and Sin-Itiro Tomonaga, the theory of Quantum Electrodynamics (QED).

While QED is a powerful theory that allows precise calculations of physical observables such as the motions of photons and electrons and the emission and absorption of a photon by an electron, the transactional interpretation is simply a way of looking at the emission and absorption of photons based on the Wheeler-Feynman attempt to describe the exchange of energy in the classical electromagnetic field as a time-symmetric process.

Wheeler-Feynman proposed adding advanced field potentials (which look like never-seen-in-nature incoming spherical waves converging on light sources) to the normal outgoing spherical waves (with retarded potentials) of classical electrodynamics. Their goal was to symmetrize electrodynamics with respect to time. There is nothing inherent in electromagnetic theory that explains the time asymmetry of radiation propagation (we see outgoing waves only).

Cramer's transactional interpretation describes an electron as sending out probabilistic "offer waves" (OW) to potential absorbers. He adds what he calls "confirmation waves" (CW) incoming to an emitter from the many possible absorbers of an emitted photon. An offer wave is not an actual photon emission, and a confirmation wave is not an actual absorption or "detection" of a photon. But Cramer did see the two waves as connecting events in spacetime. Eventually, one advanced potential confirmation wave "handshakes" with the retarded potential offer wave and produces an actual absorption.

The offer wave going out in all directions and the many confirmation waves returning are a sort of subset of the infinite number of virtual photons traveling all possible paths between emitters and absorbers in Feynman's "sum-over-paths" path-integral formulation of quantum mechanics. Kastner proposes to regard the outgoing offer wave and many incoming confirmation waves as "possible" transactions, only one of which indeterministically becomes "actual."

Kastner is a possibilist who argues that OWs and CWs are possibilities that are "real." She says that they are less real than actual empirically measurable events, but more real than an idea or concept in a person's mind. She suggests the alternate term "potentia," Aristotle's that she found Heisenberg had cited. For Kastner, the possibilities are physically real as compared to merely conceptually possible ideas that are consistent with physical law (for example, David Lewis' "possible worlds." But she says the "possibilities" described by offer and confirmation waves are "sub-empirical" and pre-spatiotemporal (i.e., they have not shown up as actual in spacetime). She calls these "incipient transactions."

Kastner describes actual spacetime events as emergent from the transaction process. She correctly thinks that Niels Bohr and Werner Heisenberg were mistaken to renounce any attempts to visualize a quantum reality underlying quantum phenomena. She describes her "new realism:"

To assume, like Bohr, that a realist understanding must be in terms of the usual "classical," causal account is to limit ourselves to a pseudo-Kantian "category of experience" which is shown to be obsolete by scientific advance, much as Kant's own prescribed "categories" became obsolete when (for example) it was discovered that theories of spacetime had to allow for non-Euclidean forms. The new realist understanding may not be in terms of causal, mechanistic processes. It may instead encompass a fundamental indeterminism at the heart of nature, but one which is well-defined in terms of the conditions under which it occurs - in contrast to prevailing "orthodox" interpretations which suffer from an ill-defined micro/ macro "cut". The new understanding offered here is a rational account, in the sense of being well-defined and self-consistent, even while it lacks certain features, such as determinism and mechanism, that have been traditionally assumed to be requirements for an acceptable scientific account of phenomena.

(The Transactional Interpretation of Quantum Mechanics, 2012, p.33)

The subtitle of Kastner's book is ""The Reality of Possibility." She says that her main thesis is that "it is perfectly reasonable to be realist about the subject matter of quantum theory" (p.28). And she calls for a new metaphysical category to describe "not quite actual...possibilities" and the Heisenberg/Aristotle idea of "potentia." 

Heisenberg took a further step in "listening" to quantum theory when he made the following statement: "Atoms and the elementary particles themselves are not real; they form a world of potentialities or possibilities rather than things of the facts."
This assertion was based on the fact that quantum systems such as atoms are generally described by quantum states with a list of possible outcomes, and yet only one of those can be realized upon measurement. I think that he was on to something here, except that I would adjust his characterization of quantum systems as follows: they are real, but not actual. In his terms, they are something not quite actual; they are "potentialities" or "possibilities." Thus my proposal is that quantum mechanics instructs us that we need a new metaphysical category: something more real than the merely abstract (or mental), but less concrete than, in Heisenberg's terms, "facts" or observable phenomena. The list of possible outcomes in the theory is just that: a list of possible ways that things could be, where only one actually becomes a "fact."

(The Transactional Interpretation of Quantum Mechanics, 2012, p.36)

Kastner claims that the transactional interpretation removes the "mystery" in John von Neumann's Process 1.

Standard quantum mechanics provides no way to "determine" when or where the wave function "collapses"

the notorious problem with the von Neumann formulation was that there seemed to be no way to determine when, why, or how the pure state should undergo such a transformation. If we take into account the physical process of absorption (i.e., state annihilation), "Process 1" becomes completely non-mysterious. It is just the process whereby the CW are returned to the emitter from all absorbers capable of responding, and a set of incipient transactions is established.

(The Transactional Interpretation of Quantum Mechanics, 2012, p.54)

She says that her "possibilities" view provides outcomes that are "clearly defined" where standard quantum theory does not.

Standard quantum mechanics prefers the basis set of eigenfunctions that corresponds to the eigenvalues (e.g., pointer readings, detector states) of the measuring apparatus and the target system

If we adopt the approach that quantum theory tells us about many possibilities arising from interactions between offer waves and confirmation waves, then we gain a clearly defined set of possible outcomes missing in the standard account, which disregards the real physical process of absorption. Recall that one component of the measurement problem is the amplification of the quantum state through interactions with the measuring apparatus, the first observer, the second observer, etc., with no means of deciding when the measurement has been completed. The designation of a stage at which the measurement is "completed" is referred to as the "Heisenberg cut," which is notoriously arbitrary. The arbitrariness is removed once we notice that the original offer wave inevitably encounters one or more absorbers that generate confirmations in response to the offer. It is at that point that set of incipient transactions is established.

(The Transactional Interpretation of Quantum Mechanics, 2012, pp.202-203)

The information interpretation of quantum mechanics identifies the Heisenberg/von Neumann "cut" unambiguously as the irreversible creation of stable information in the world (e.g., a detector click or spot on a photographic plate) that may later be observed and constitute a "measurement." For the location of the "cut," see John Bell's possible locations for the shifty split"

Many interpretations of quantum mechanics, especially those popular today, say that deterministic evolution of the Schrödinger wave equation makes the theory "unitary." "Unitarity" conserves many expectation values, including the total information at all times. These interpretations deny the "collapse" of the wave function and the appearance of a particle somewhere. Many say that there are no particles, only fields. She writes...

It has been argued that if the non-unitary measurement transition of Von Neumann is a physically real component of quantum theory, then the representation of the system(s) under study by proper mixed states, subject to a probabilistic master equation description relative to a distinguished basis, becomes physically justified. This rectifies a weakness in the usual approach, which helps itself to the convenient basis and accompanying probabilistic description (effectively Pauli’s “random phase assumption”) as a “for all practical purposes” approximation. However, the utility of a probabilistic expression for calculational purposes does not constitute theoretical justification for the probabilistic description, which is needed in order for the “coarse-graining” and resulting entropy increase to describe what is physically occurring in a system. Once we have that justification, through real non-unitary collapse, we have the microscopic irreversibility needed to place the H-theorem on sound physical footing.

According to the TI account of measurement, a quantum system undergoes a real, physical non-unitary state transition based on absorber response, which projects it into a Boolean probability space defined with respect to the observable being measured (typically energy in the context of thermodynamics). Thus the system’s probabilistic description by random variables is justified; the non-unitary measurement transition can be understood as the physical origin of the “initial probability assumption” referred to as puzzling by Sklar. In this model, it ceases to be an assumption and can be seen as describing a physical feature of Nature. In effect, Boltzmann was completely correct about the Stosszahlansatz, even though he could not explain why in classical terms.

In addition, the relativistic level of TI (referred to as RTI) provides a physical reason for the directionality of the irreversibility inherent in the measurement transition at the micro-level, thereby establishing an arrow of time underlying the Second Law. In this respect, the microscopic arrow of time becomes a component of the explanation for the increase in the entropy of closed systems towards what we call “the future”, without the need for an additional postulate of a cosmological low-entropy past.

The present model has been contrasted with the GRW model proposed by Albert, as follows: it does not require any change to the basic Schrödinger evolution, but simply provides a physical account of the non-unitary measurement transition previously formalized by Von Neumann. In addition, the present model takes conserved physical quantities (energy, momentum, etc.) as the preferred eigenbasis of collapse, rather than position as in the GRW theory. This takes into account the fact that position is not an observable at the relativistic level (and time is not an observable at any level). It is also in accordance with the naturally occurring microstates of thermodynamical systems, in which the molecular components are described by distributions over their energies and momenta (i.e., the Boltzmann distribution applies to energies, not positions).

(On Quantum Collapse as a Basis for the Second Law of Thermodynamics, p.10)