{"id":"1063315c-9aac-4874-a8aa-98fc42ed1cba","arxiv_id":"1908.03281","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A liquidity taker's latency-optimal discretion is characterized by δ*_t = 2γE_{t-}[D_T]+γ+α, the solution of a random-measure FBSDE, but the optimality proof and numerical regime contain gaps.","lead":"This paper derives the optimal price limit a trader should attach to a marketable order when network delay makes the order book a moving target. The optimal discretion to walk the book is the solution of a new forward-backward stochastic differential equation, with numerical examples showing lower expected cost than fixed-discretion orders.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Global optimality of δ* is unproven: Theorem 6's contradiction argument is invalid and Theorem 5's second-derivative condition is mis-stated as a bilinear form; the central claim needs a convexity/existence argument.","rationale":"The reader's conditional verdict is appropriate, and the identified weak spot is real, but the most load-bearing issue is broader than the asserted non-negativity of (21b). The proof of global optimality has two independent gaps: the second-derivative criterion is mis-stated (and possibly missing a jump term), and Theorem 6's contradiction argument cannot establish that δ* is a global minimizer without an existence argument and a uniqueness result that is only proven under a smallness condition violated by the paper's own numerics. I partially agree with the reader because the second-derivative concern is the right area, but the specific claim that the bilinear form is negative for oscillatory perturbations when γ is large is not the correct objection: a quadratic form with w=ν is non-negative for all γ. The real issue is that the proof as written does not show local optimality either, because the second derivative is not correctly reduced to the quadratic form and the D_{t−} derivative term appears to be omitted. The first-order condition and FBSDE machinery are valuable, and the result may be repairable by adding a strict-positivity assumption on φ, a convexity argument, or an explicit proof of existence of a global minimizer; hence a conditional acceptance with these required fixes remains the right verdict rather than rejection. No code or data is supplied, which further supports keeping the verdict conditional.","tokens_in":21977,"tokens_out":17362,"duration_ms":187973,"concrete_test":"Independently re-derive Eq. (21b) from the definition of the second Gateaux derivative, keeping the derivative of D_{t−} and setting w=ν; evaluate the resulting quadratic form for a two-jump Poisson example with marks uniform on [0,1] and γ=10. If the derived form differs from (21b) or is negative, the local-optimality proof fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that (12) gives the global optimum rests on Theorems 5 and 6. Theorem 5 asserts that the second Gateaux derivative (21) is non-negative for every ν,w∈A. This is false as written: for fixed δ, (21b) is linear in w, so if it is non-zero for some ν, taking w=−ν gives a negative value. The local-minimum condition requires the quadratic form ⟨D²J(δ),w,w⟩≥0, not bilinear non-negativity. Even then, (21b) appears to miss the derivative of D_{t−} w.r.t. δ, which contributes a term involving the past jumps of q=p−p̃; that term is not shown to be non-negative. Theorem 6 is also not a valid global-optimality proof. Lemma 4 plus linearity of the directional derivative in w already forces ⟨DJ(δ̂),w⟩=0 for all w (apply it to −w). The subsequent claim that δ̂≠δ* implies existence of ŵ with positive derivative is unsupported; a flat region of φ (e.g., marks uniform on [0,1], δ≥1) gives zero derivative for all directions while δ̂≠δ*. Identifying δ̂ with δ* would require uniqueness of the FBSDE solution (Theorem 4, only under kTλ̄(max{1,2γ})²<1) and an existence proof for a global minimizer, neither of which is supplied. Numerical runs (λ=100, γ=0.1) lie far outside the theorem's contraction condition.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper models the latency faced by a liquidity taker who sends marketable limit orders (MLOs) to a moving limit order book. The interaction is described by a marked point process, and the agent chooses a predictable 'discretion' δ that determines how far the MLO may walk the book. The performance criterion penalizes expected walking costs, a linear penalty α on missed trades, and a quadratic penalty γ on missed trades. Using Gateaux derivatives, the authors derive a first-order condition characterizing a candidate optimal discretion δ* as the solution of the FBSDE δ*_t = 2γ E_{t-}[D^{δ*}_T] + γ + α (Eq. (12)). They prove existence and uniqueness of this FBSDE under a contraction condition, claim that δ* is the global minimizer of the performance criterion, and implement the strategy numerically via a PIDE for the value function.","tokens_in":22348,"tokens_out":41775,"duration_ms":449979,"significance":"If the optimality claim were rigorously established, the paper would provide a useful, explicit framework for latency-aware execution: the γ=0 case gives a closed-form constant discretion, and the γ>0 case gives a tractable FBSDE/PIDE characterization. The modelling choices are clear and the FBSDE formulation is genuinely new to the execution literature. The paper also makes falsifiable numerical predictions about fill ratios and walking costs. However, the central proof that δ* is the global optimum is not valid, and the numerical experiments lie outside the proven uniqueness regime. The framework is promising, but the overclaim that δ* is the argmin is currently unsupported.","major_comments":[{"comment":"The contraction estimate in Theorem 4 undercounts the Lipschitz constant. From the two bounds displayed after Eq. (20), the coefficient on ||U−X|| in the sum is k λ̄ T (1+2γ), not k λ̄ T max{1,2γ}², and the coefficient on ||V−Y|| is 2γ(1+2γ)kλ̄T. The contraction condition should therefore involve (1+2γ)max{1,2γ} k λ̄ T < 1. Moreover, the numerical experiments in §5 use λ=100, T=1, and a normal mark distribution for which k≈0.4, so kTλ̄≈40; even the paper's stated condition is violated by two orders of magnitude. The numerical section thus lies outside the proven existence and uniqueness regime.","section":"§2.3 and §4, Theorem 6"}],"minor_comments":[{"comment":"Lemma 3 assumes a finite bound ar N on the number of trade attempts that is not part of Assumption 1; the lemma should either be restricted to models with N_T≤ar N or proved without such a bound. The final display of the proof also uses an undefined symbol ar{z}, which appears to be a typo for α.","section":"§4, Lemma 3"},{"comment":"Theorem 3 states uniqueness of the fixed point of Θ, but its proof only establishes convergence of the Picard iterates to a fixed point; a separate uniqueness argument is needed.","section":"§3, Theorem 3"},{"comment":"The assertion that existence and uniqueness of the PIDE follow from a comparison principle is not backed by a stated or proved comparison theorem for this nonlinear, nonlocal equation. Since the numerical strategy is obtained by solving this PIDE, this gap should be addressed or the PIDE solution should be presented as a heuristic approximation.","section":"§5, Theorem 7"},{"comment":"There are several typographical issues, including 'Lipchitz' for 'Lipschitz' in Assumption 1 and Theorem 4, and the phrase 'almost everywhere in T×Ω' in Theorem 1 should be specified with respect to the product measure dA⊗dP.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The modelling idea and the FBSDE/PIDE framework are potentially valuable, but the paper currently overclaims optimality. I recommend major revision rather than rejection because the core framework is salvageable: the authors should add explicit assumptions (e.g., positive mean marks) that ensure a minimizer exists, replace the variational global-optimality proof with a verification theorem or a correct convexity argument, and correct the contraction constant in Theorem 4. The numerical section should also be reconciled with the proven parameter regime."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Colleague],\n\nYou asked about the Cartea-Jaimungal-Sánchez-Betancourt latency paper. My quick take: the economic model and the FBSDE characterization are genuinely new, but the proof that the candidate control is globally optimal does not survive contact with the paper. I'd still send it out, but the referee will need to send it back for serious work.\n\nWhat's actually good: they set up a marked point process for LOB updates against a latency-delayed marketable order, and a performance criterion that balances walking-the-book cost against linear and quadratic penalties for missed trades. The variational derivation of the first-order condition is clean, and the resulting representation δ*_t = 2γ E_{t-}[D_T] + γ + α is a reasonable candidate. For γ=0 you get the constant α, which is nice and usable. The existence and uniqueness result for the random-measure FBSDE (Theorem 4) looks correct under the stated contraction condition.\n\nThe soft spots are real. Theorem 5 asserts that the second Gateaux derivative is non-negative for every pair of directions; the expression (21b) is bilinear, not a quadratic form, so the local-minimality claim is false as written—taking w = -ν gives the opposite sign. Even the quadratic form with w=ν isn't obviously non-negative, and there's a potential missing term from the dependence of D_{t-} on δ. Theorem 6 doesn't fix it. Once you combine Lemma 4 with w and -w, the derivative at a global minimum must vanish in every direction; then the alleged \"there exists w with positive derivative\" is inconsistent. What's missing is an existence proof for a global minimizer, which is not supplied. Also, Theorem 1's \"if and only if\" requires φ_t(δ_t)>0; with a density that vanishes on an interval (e.g., uniform marks with δ>1), the derivative vanishes even when δ is far from δ*. The numerics run with λ=100, γ=0.1, which is far outside the contraction bound kTλ̄(max{1,2γ})²<1.\n\nWho gets value from this? People working on latency models in market microstructure will find the setup and the FBSDE angle worth reading. The optimality story is not yet established, so I wouldn't cite the optimality results as theorems until fixed.\n\nMy recommendation: send to peer review. It's exactly the kind of paper a serious referee should engage with—the model is worthwhile, the mathematical issues are identifiable and fixable, and the authors should be asked to prove strict positivity assumptions, establish existence of minimizer, and rerun numerics inside the contraction regime.","headline":"New FBSDE characterization of latency-optimal price limits, but the global optimality proof is invalid; deserves review with major revision.","tokens_in":22832,"tokens_out":5725,"would_cite":false,"duration_ms":59410,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["91G80","60G55","60H10","49K45"],"pacs":[],"model":"deepseek-v4-flash","headline":"Optimal discretion to walk the book is characterized by a forward-backward SDE driven by random measures, and the paper proves existence, uniqueness, and global optimality of the solution.","keywords":["latency","marketable limit orders","walking the book","marked point processes","forward-backward stochastic differential equations","limit order book","optimal execution","fill ratio"],"falsifier":"Compute the second Gateaux derivative (21a)-(21b) at $\\delta^*$ for a smooth mark density $\\varphi$ and high-frequency perturbations $\\nu_t = w_t = \\sin(n t)$ with large $n$ and sufficiently large $\\gamma$; if the bilinear form is negative, $\\delta^*$ is not a local minimum, contradicting Theorem 5. Alternatively, run the paper's 10,000-simulation design with a miss penalty $\\gamma$ above the range tested and compare the empirical $J$ of $\\delta^*$ against a nearby constant-discretion policy: a lower $J$ for the constant policy would falsify global optimality.","tokens_in":21791,"feed_emoji":"📈","tokens_out":5073,"duration_ms":52572,"temperature":0.7,"pith_summary":"The paper tries to show that a liquidity taker whose orders reach the exchange after a delay can choose, for every order, exactly how much extra price to allow (the 'discretion to walk the book') to balance the cost of worse fills against the cost of missing trades. It models the limit order book as a moving target driven by a marked point process and derives the optimal discretion as the solution to a forward-backward stochastic differential equation driven by random measures. When the penalty for missed trades is linear, the optimal discretion is constant; when it is quadratic, the optimal discretion grows with the cumulative number of misses. If this is right, a trader who knows the arrival intensity of her orders and the distribution of book shocks can compute the optimal price limit path in closed form in the linear case and numerically otherwise. The paper positions this as a layer that any existing liquidity-taking strategy could add, with applications beyond equity trading, such as foreign-exchange last-look market making.","feed_headline":"One formula sets how deep to walk the book","feed_subtitle":"Latency-optimal orders balance missed trades against walking costs; quadratic penalties make discretion respond to past misses.","key_machinery":"The machinery is a marked point process $N = \\{(T_n, Z_n)\\}$ whose compensator has intensity $\\lambda$ and mark distribution $\\varphi$; a fill occurs when the book shock $z$ does not exceed the chosen discretion $\\delta$, so the controlled cost and miss processes are $C_t^\\delta = \\int_0^t\\int_{\\mathbb{R}} z \\hat{G}(\\delta_s - z)\\,p(dz,ds)$ and $D_t^\\delta = \\int_0^t\\int_{\\mathbb{R}} G(\\delta_s - z)\\,p(dz,ds)$. Taking Gateaux derivatives of $J$ leads to the optimality equation $\\delta^*_t = 2\\gamma \\mathbb{E}_{t-}[D_T^{\\delta^*}] + \\gamma + \\alpha$, which is recast as a new class of random-measure-driven FBSDE; existence and uniqueness are obtained by fixed-point arguments (contraction for the backward part, Borel-Cantelli for the forward part, and a joint contraction for the full system), and under a Markovian intensity assumption the control is characterized by a partial integro-differential equation for $h(t,D,\\lambda)$.","core_discovery":"The central claim is that for a performance criterion $J(\\delta) = \\mathbb{E}[C_T^\\delta + \\alpha D_T^\\delta + \\gamma (D_T^\\delta)^2]$, the optimal discretion $\\delta^*$ satisfies $\\delta^*_t = 2\\gamma \\mathbb{E}_{t-}[D_T^{\\delta^*}] + \\gamma + \\alpha$ (Equation 12), where $D$ is the count of missed trades. The authors prove existence and uniqueness of the solution to this FBSDE, prove that any global minimum of $J$ coincides with $\\delta^*$, and for $\\gamma=0$ obtain the explicit fixed-discretion rule $\\delta^* = \\alpha$. Numerically, they show the quadratic-penalty strategy achieves a given fill-probability target at lower expected cost than the fixed-discretion strategy.","pith_inferences":["The paper leaves open whether the second-order condition actually holds globally; a direct verification for high-frequency perturbations would either close the proof or restrict the admissible set.","Because $\\delta^*$ is a càglàd sub-martingale, the optimal rule resembles a state-dependent control limit, which suggests a connection to optimal stopping or barrier policies for order execution.","A testable extension is to estimate $\\varphi_t$ nonparametrically from limit-order-book snapshots and compare realized fill probabilities and costs of $\\delta^*$ versus fixed discretion on out-of-sample data.","The FBSDE formulation suggests that similar latency problems, such as queue position, partial fills, or multiple venues, would produce FBSDEs of the same form and could be solved with the same variational machinery."],"forward_implications":["A trader with estimates of her order-arrival intensity and the shock distribution can compute the optimal discretion path with no simulation in the $\\gamma=0$ case and with a numerical PIDE solve otherwise.","The optimal discretion increases with accumulated misses when $\\gamma>0$, so the strategy automatically opens the book after a string of unfilled orders and tightens it after fills.","Under the calibrated parameters, a quadratic-penalty strategy reaches a 95% probability of missing fewer than 10% of attempts at lower expected cost (about 5.93) than the best fixed-discretion strategy (about 9.52).","At the cost-neutral parameter $\\gamma \\approx 0.0693$, the expected cost of filled trades is zero while the miss rate is about 10.5%, giving a default parameter choice for a trader without a miss penalty.","The same framework gives a foreign-exchange market maker under last look a rule for how much adverse price movement to tolerate before rejecting an incoming order."],"supporting_citations":[{"why":"Supplies the marked-point-process and random-measure BSDE framework used to set up the sample space and compensator.","marker":"(Confortola et al. 2016)"},{"why":"Provides the predictable-projection theorem used to identify processes with their left-continuous versions.","marker":"(Cohen and Elliott 2015)"},{"why":"Provides the contraction fixed-point argument that underpins the existence proof for the backward part of the FBSDE.","marker":"(Duffie and Epstein 1992)"},{"why":"Gives the empirical fill-ratio and no-discretion baseline that the numerical comparison outperforms.","marker":"(Cartea and Sánchez-Betancourt 2018)"},{"why":"Establishes the cost-of-latency benchmark that motivates the model of delayed marketable orders.","marker":"(Moallemi and Sağlam 2013)"},{"why":"Provides an alternative execution strategy under latency that the paper contrasts with its own optimal-discretion approach.","marker":"(Stoikov and Waeber 2016)"}],"fun_headline_variants":["Latency-optimal orders tune how deep to walk the book","New FBSDE guides marketable order price limits","Strike the balance: missing fewer trades without overpaying","Optimal order depth: a formula for the walking cost tradeoff","How deep to go? A latency-aware strategy says: it depends"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof that the candidate $\\delta^*$ is a local minimum assumes the second Gateaux derivative of $J$ is nonnegative in every admissible direction, and this is asserted rather than proved; with large quadratic miss penalty and oscillatory perturbations the bilinear form can be negative, so the local and global optimality of $\\delta^*$ is not fully established.","fun_headline_variants_meta":{"raw":{"variants":["Latency-optimal orders tune how deep to walk the book","New FBSDE guides marketable order price limits","Strike the balance: missing fewer trades without overpaying","Optimal order depth: a formula for the walking cost tradeoff","How deep to go? A latency-aware strategy says: it depends"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000466,"raw_usage":{"total_tokens":2313,"prompt_tokens":921,"completion_tokens":1392,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":537,"completion_tokens_details":{"reasoning_tokens":1306}},"tokens_in":537,"tokens_out":1392,"duration_ms":12963,"temperature":1.0,"reasoning_tokens":1306,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:21:04.831584+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the second Gateaux derivative (21a)-(21b) at $\\delta^*$ for a smooth mark density $\\varphi$ and high-frequency perturbations $\\nu_t = w_t = \\sin(n t)$ with large $n$ and sufficiently large $\\gamma$; if the bilinear form is negative, $\\delta^*$ is not a local minimum, contradicting Theorem 5. Alternatively, run the paper's 10,000-simulation design with a miss penalty $\\gamma$ above the range tested and compare the empirical $J$ of $\\delta^*$ against a nearby constant-discretion policy: a lower $J$ for the constant policy would falsify global optimality.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the marked-point-process and random-measure BSDE framework used to set up the sample space and compensator."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the predictable-projection theorem used to identify processes with their left-continuous versions."},{"cited_title":"and Epstein, L","cited_arxiv_id":null,"evidence_quote":"Provides the contraction fixed-point argument that underpins the existence proof for the backward part of the FBSDE."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the cost-of-latency benchmark that motivates the model of delayed marketable orders."},{"cited_title":"and Waeber, R","cited_arxiv_id":null,"evidence_quote":"Provides an alternative execution strategy under latency that the paper contrasts with its own optimal-discretion approach."}],"review_version":1}